A comparison of biochemical pregnancy rates between women who underwent IVF and fertile controls who conceived spontaneously^†

Abstract

STUDY QUESTION

Does IVF affect the biochemical pregnancy rate?

SUMMARY ANSWER

The likelihood of an early pregnancy loss may be lower and is certainly not higher in IVF cycles when compared with published rates of biochemical pregnancy in fertile women ≤42 years old.

WHAT IS KNOWN ALREADY

The use of gonadotrophins to stimulate multi-folliculogenesis alters endometrial expression of genes and proteins, compared with unstimulated cycles. Exogenous estrogen and progesterone taken for endometrial preparation in frozen embryo transfer cycles, also cause changes in endometrial gene and protein expression .These endometrial alterations may compromise the ability of embryos to develop once implanted, possibly increasing the biochemical pregnancy rate.

STUDY DESIGN, SIZE, DURATION

This is a retrospective study, involving 1636 fresh and 188 frozen, single embryo transfer (SET) IVF cycles performed between August 2008 and December 2012. The biochemical pregnancy rate of the 1824 combined IVF and frozen cycles were compared with fertile controls, derived from the three prospective studies in the medical literature that evaluate this rate.

PARTICIPANTS/MATERIALS, SETTING, METHODS

Subjects ≤42-years old, who underwent a SET, as part of a fresh or thawed IVF cycle were considered for inclusion. Each subject is represented only once. The biochemical pregnancy rates were compared with those of historical standard, fertile populations with spontaneous conceptions.

MAIN RESULTS AND THE ROLE OF CHANCE

The pregnancy rates per transfer for fresh and frozen IVF cycles were similar at 39 and 40%, respectively. There was also no significant difference in the likelihood of pregnancy outcomes (clinical, biochemical and ectopic pregnancy) between fresh IVF and frozen cycles (85.4 versus 85.6%, 13.8 versus 14.8%, 0.5 versus 0%, P = 0.82). However, pregnancy rates decreased in older patients when compared with younger ones P < 0.0001. The biochemical pregnancy rate for fresh and frozen IVF cycles combined was 13.8% of all pregnancies. IVF and frozen cycles were combined as the IVF group treated with hormones for further comparison with the fertile control group. The biochemical pregnancy rate (14%) in the IVF group was lower than the rate based on the total fertile group (18%), P = 0.01 and differed significantly from the rate in two out of the three studies used to establish the normative rate. The age ranges of the IVF and fertile controls were 21–42 years. The mean age in the IVF population was 34.8 years, as compared with 29 years, 29, 4 years and 30.6 years (Zinaman) in the three published studies (mean: 29.4 years).

LIMITATIONS, REASONS FOR CAUTION

This is a retrospective study and it was impossible to recruit an in-house biochemical pregnancy control population.

WIDER IMPLICATIONS OF THE FINDINGS

Lower early pregnancy wastage after IVF may be due to the opportunity to select the embryo for transfer. This finding should be confirmed in further studies but supports the idea that embryo selection is an important step.

STUDY FUNDING/COMPETING INTEREST(S)

None.

Introduction

Serum and urinary β-human chorionic gonadotrophin (β-hCG) concentrations are used to detect pregnancy within 13 and 11 days after, Day 3 embryo or blastocyst transfer, respectively. One and 2 weeks after the positive pregnancy test, a transvaginal ultrasound (TVUS) can confirm an intrauterine gestational sac and a fetal pole, respectively. The presence of an intrauterine gestational sac is defined as a clinical pregnancy. In 18–22% of pregnancies, the TVUS failed to confirm the existence of a gestational sac. This may be an early pregnancy loss or occult pregnancy (Alfredsson, 1988; Wilcox et al., 1988; Kolstad et al., 1999). A positive β-hCG test in either serum or urine without the development of a gestational sac is also known as a biochemical pregnancy, unless an extra uterine (ectopic) pregnancy is present. In biochemical pregnancies, β-hCG levels usually decline without any treatment.

The use of gonadotrophins to stimulate multi-folliculogenesis alters endometrial expression of genes and proteins, when compared with unstimulated cycles (Pilka et al., 2006; Horcajadas et al., 2008a, b; Haouzi et al., 2010). In patients undergoing controlled ovarian stimulation (COS) when compared with natural cycles, experiments using immunohistochemical and histological markers of the implantation window have demonstrated endometrial alteration during the luteal phase (Bourgain and Devroey, 2003). The morphological progressions in the early luteal phase of the endometrium in COS cycles has been investigated by scanning microscopy studies (Giudice, 2003; Papanikolaou et al., 2005). Some studies advocate that ovarian stimulation diminishes uterine receptivity when compared with natural cycles (Simon et al., 1995; Fauser and Devroey, 2003). The deleterious effect of gonadotrophin stimulation on uterine receptivity has been reported to be worse in high-responder patients comparing with average responders (Pellicer et al., 1996; Simon et al., 1998; Simon et al., 2003; Murata et al., 2005). Recent results with frozen thawed embryo transfer cycles have suggested higher pregnancy rates than in fresh autologous cycles (Psychoyos, 1986; Shapiro et al., 2011), likely due to a decreased effect on endometrial receptivity. In a prospective study, it was recently reported that the large increases in serum estradiol concentrations and progesterone levels in the late follicular phase lead to a modified steroid receptor profile. This late follicular profile resembles that of the early luteal phase. These alterations can then lead to accelerated differentiation of the endometrium as well an earlier presence of pinopodes (Develioglu et al., 1999). Down regulation of endometrial estradiol and progesterone receptors have also been demonstrated in immunohistochemical studies performed on COS cycles (Simon et al., 1996). In addition, biochemical changes (gene expression) in the endometrium were reported in COS when compared with natural cycles (Horcajadas et al., 2005). When comparing COS for IVF with natural cycles, in the same patient, it has been shown that a significant effect on the endometrial gene expression occurs during the window of implantation (Paulson et al., 1990; Simon et al., 2005).

Pinopodes are acknowledged as an accurate marker for endometrial receptivity (Nikas, 1999). Scanning electron microscopy allows visualization of the remarkable modifications of the endometrial epithelium around the period of implantation. In unstimulated cycles, the epithelial cells lining the apical membranes of the uterine cavity lose their microvilli and develop large and smooth membrane projections during the implantation window (Develioglu et al., 1999). About 1 week after ovulation, the pinopodes are expressed and then regress within just 2 days. However, in COS cycles pinopodes formation occurs earlier whereas in hormone replacement therapy prepared frozen embryo transfer (FET) cycles the pinopods occur later, when compared with natural cycles (Nikas, 2000; Nikas and Aghajanova, 2002; Nikas and Makrigiannakis, 2003; Pilka et al., 2006). The previously listed hormone and gene alterations seen in these stimulated cycles have been postulated to be the mechanism of changes in pinopod expression. Ultimately, these alterations in pinopod expression may explain the lower probability that fresh embryos implant when compared with frozen thawed embryos. This is essentially due to a more pronounced desynchronization of the implantation window in COS cycles, than in estrogen and progesterone FET cycles.

It could be hypothesized that the endometrial alterations seen in COS and FET cycles may compromise the ability of embryos to develop once implanted, possibly increasing the biochemical pregnancy rate. This hypothesis is supported by the findings of Wang, who found that subfertile couples have higher miscarriage rates after fertility treatment than spontaneous conception (Wang et al., 2004). It was only recently, with the wide use of single embryo transfer (SET), that the possibility of an undetected chemical pregnancy occurring concurrently with a clinical pregnancy could be mitigated in IVF cycles. Therefore, the effect of endometrium embryo interaction on biochemical pregnancy rates can now be studied in isolation.

The aim of this study is to investigate the effect of in vitro fertilization on the biochemical pregnancy rate and to compare the rates of biochemical pregnancy due to frozen and fresh embryo transfer cycles, since they alter pinopod expression in different ways. A comparison with rates of biochemical pregnancy from the IVF population will be contrasted with the rate reported in the fertile population. A review of the medical literature suggests this is the first study to investigate the biochemical pregnancy rate after IVF.

Methods

A retrospective study was performed, which evaluated subjects who underwent SET after gonadotrophin stimulation as part of a fresh IVF cycle or as part of an exogenous hormone supported FET. Subjects were enrolled in the study if they underwent treatments between August 2008 and December 2012. All subjects who were <43 years old, and underwent a SET, as part of a fresh or thawed IVF cycle were considered for inclusion. Exclusion criteria were egg donation, patient who had recurrent implantation failure, recurrent pregnancy loss, pre-implantation genetic screening/pre-implantation genetic diagnosis and patients after social or medical fertility preservation. Subjects with social fertility preservation were excluded because they are not infertile. Those with medical preservation besides not being infertile, often have sub adequate COS due to the rapid need to begin chemotherapy therefore, they were excluded.

Each subject is represented only once in the data base. The indications for IVF were polycystic ovary syndrome, male factor, tubal factor, endometriosis, poor ovarian reserve and unexplained infertility. The biochemical pregnancy rates for these two groups were compared with those of historical standard, fertile populations with spontaneous conception. The biochemical pregnancy rate for the fertile population was determined using the three prospective studies in the medical literature that evaluate this rate (Wilcox et al., 1988; Ellish et al., 1996; Zinaman et al., 1996). A PubMed search failed to find any other studies in the medical literature which calculated a biochemical pregnancy rate for the fertile population. The mean biochemical pregnancy rate was also calculated for these three studies combined and compared with the rate in the IVF patients.

Patients had a quantitative serum β-hCG level performed 11 days after embryo transfer (blastocyst) or 13 days after Day 3 (cleavage embryo) transfers. A pregnancy test was defined as positive if serum β-hCG levels were ≥10 mIU/ml using the Siemens Immulite 2000 assay. Sensitivity of the assay was 0.4 mIU/ml, with <1% cross-reactivity. Intra- and inter-assay coefficients of variability were <6.6 and 7.4%, respectively.

Biochemical pregnancy was defined as a positive pregnancy test in the absence of any ultrasonographic evidence of pregnancy, and no evidence or treatment of an extra uterine pregnancy. The TVUS was performed 2 weeks after a positive pregnancy test.

For the three studies used as a historical standard, their methods of determining pregnancy are henceforth described. Wilcox measured urinary concentration of hCG with the use of an immunoradiometric assay that was able to detect levels as low as 0.01 ng per milliliter with 100% specificity for hCG in the presence of LH hormone. Pregnancy was defined if the level of hCG was >0.025 ng/ml on 3 consecutive days. In this study, urinary hCG levels were measured by means of an immunoradiometric assay with a detection antibody (R525) that reacts with the unique carboxyterminal region of hCG beta-chain ‘This property of the R525 antibody allows HCG to be measured with no cross-reaction with LH. The assay measures intact HCG with same detection of free beta subunit of HCG (Wilcox et al., 1988)’.

The two other studies performed by Ellish et al., (1996) and Zinaman et al., (1996) performed similar detection techniques. They had subjects collect and freeze daily morning urine samples. Retrospectively, urinary hCG was measured starting 9 days after suspected ovulation. hCG was tested using a monoclonal antibody sandwich assay which could detect complete and fragmented hCG (Dynatech immulon 2 microtiter plates). The assay sensitivity was in the range of 0.06 ng/ml, the intra- and inter-assay coefficients of variation were 4.6 and 10.2%, respectively. An occult pregnancy was considered to have occurred when hCG titers exceeded 0.15 ng/ml urine for greater than 3 consecutive days. All three studies used the disappearance of urinary hCG as a criterion of a biochemical pregnancy. Zinaman and Ellish both added the absence of ultrasound evidence of an intrauterine or extra-uterine pregnancy but failed to specify when the ultrasound was done. Wilcox relied on the diagnosis of the woman's physician and also failed to specify when the ultrasound examination was done.

Biochemical pregnancy rates for the fertile subjects was also obtained by calculating a mean rate for the studies in the medical literature. The mean standard rate for biochemical pregnancy in the fertile population was based on 429 subjects. This rate for the three studies combined was determined by dividing the total number of biochemical pregnancies in these studies by the total number of subjects evaluated, by the three studies.

In vitro fertilization

Patients who underwent COS for IVF were treated under three different protocols: the long-gonadotrophin releasing hormone agonist (GnRH-ag) protocol, the short fix GnRH antagonist protocol, and the microdose-flare protocol, as previously described (Oron et al., 2014).

Mature oocytes were inseminated using the partner's spermatozoa. The zygotes were cultured in COOK cleavage medium (Cook Medical, Bloomington, IN, USA). In the McGill University Health Center, reproductive center the same culture medium was used between 2008 and 2012.

The embryo transfer was performed 2–5 days after oocyte collection, depending on the number and quality of the embryos available. The luteal phase was supplemented with 17β-estradiol and progesterone starting from the day of oocyte collection up to the day of the pregnancy test and throughout the first trimester, if pregnant.

Frozen embryo transfer cycles

The FET was performed after treatment of the recipients with 17β-estradiol starting on Day 2 or 3 of the menstrual cycle for at least 15 days before embryo transfer until the (trilaminar) endometrial thickness reached at least 8 mm (measured by TVUS), Embryos were accepted for transfer if they retained ≥50% of blastomeres intact after thawing.

Only one embryo was transferred in a manner similar to the fresh IVF cycle. The luteal phase was created with vaginal or i.m. progesterone. The progesterone supplementation was started before transfer in as many days as the age of the embryo to be transferred plus one, up to the day of the pregnancy test and throughout the first trimester, if pregnant. Most of embryos at our institution are frozen as blastocysts (Bucket et al., 2008; Oron et al., 2014).

Statistical analysis

Statistical analysis was performed using the statistical package for social sciences (SPSS version 20). Data were compared using χ² contingency and goodness of fit tests. A stepwise logistic regression analysis was performed to determine risk factors for biochemical pregnancy among women who conceived. A two sided P-value of ≤0.05 was accepted as statistically significant. Data are presented as the mean ± standard deviation.

Results

The study included 1824 subjects (each subject was only included once), 1636 fresh IVF cycles (1006 not pregnant, 538 clinical pregnancies, 87 biochemical pregnancies and 5 ectopic pregnancies) and 188 FET cycles (112 not pregnant, 65 clinical pregnancies, 11 biochemical pregnancies and 0 ectopic pregnancies) SETs. The mean rate of biochemical pregnancy in the fertile control population was based on 429 subjects (Table I).

The age range of the IVF population was 21–42 years. The mean age in the IVF population was 34.8 ± 4.4 years, when compared with 29 years (Wilcox), 29.4 years (Ellish), 30.6 ± 3.3 years (Zinaman) and 29.4 years (mean of 3 studies). The standard deviation was not provided by Wilcox and Ellish in their studies.

The age range for the Wilcox and the Ellish studies were not presented in the respective articles. The age range for the Zinaman study was 23–37 years. The fertile population is younger than the infertile group (29.4 versus 34 years) on average, although P-values could not be calculated because standard deviations were not provided in two of the control studies.

The pregnancy rates per transfer for fresh and frozen IVF cycles were similar at 39 and 40%, respectively. The probability of different pregnancy outcomes (no pregnancy, clinical pregnancy, biochemical pregnancy and ectopic pregnancy) did not differ when comparing fresh and frozen IVF cycles, P = 0.82. The biochemical pregnancy rate for fresh and frozen IVF cycles was 14% of all pregnancies, for both groups. Since there was no difference in the rate of biochemical pregnancies in these two groups, who underwent reproductive technology to conceive, they were combined for further analysis.

As expected, pregnancy rates decreased in older patients when compared with younger ones P < 0.0001.

The pregnancy outcomes for the IVF group and the three studies used as standards for the biochemical pregnancy rate in the fertile population is presented in Table II.

The biochemical pregnancy rate was 14% in the IVF group. This rate was lower than the rate based on the total fertile group which was 18%, P = 0.0079, and differed statistically from the rate in two of the three studies used to establish the normative rate.

Stepwise logistic regression analysis was performed to determine risk factors for biochemical pregnancy among women who conceived. The results are presented in Table III. No factors reached statistical significance, except for one. Biochemical pregnancy rates were lower when more blastocysts were transferred than cleavage stage embryos.

Discussion and Conclusion

The results of this study show that early pregnancy wastage may be lower in subjects who have embryo transfers when compared with spontaneous pregnancies. This finding is surprising, given the expected alterations in the endometrium with exogenous gonadotrophins or estrogens and progesterone taken for a FET. It was expected that the biochemical pregnancy rate would be higher or similar to spontaneous conceptions, in women who underwent reproductive technology. This discovery may be related to the selection of the best embryo being transferred when compared with single embryo genesis in nature. Clearly, this find was not due to a younger age in the IVF group since the mean age in all three groups used to establish the standard in the fertile population was likely younger.

This finding is also not likely compromised by the spontaneously resolving ectopic pregnancy rate. What we call the biochemical pregnancy rate was defined in the same manner in the three studies performed in the fertile population, and are actually pregnancies of unknown location. Some of these pregnancies may be extra-uterine ones, which resolved spontaneously on their own. However, the ectopic pregnancy rate should be highest in the infertile not the fertile population, particularly since patients with tubal diseases were included among the infertile subjects. Therefore, the true biochemical and the ectopic pregnancy rate combined should have been higher not lower in the infertile group.

A PubMed search (early pregnancy loss, fertile population, human fertility and biochemical pregnancy) detected that there are only the three cited studies in the medical literature, which have determined a biochemical pregnancy rate in the fertile population. The mean biochemical pregnancy rate for the fertile subjects were was based on 429 subjects. The biochemical pregnancy rate in each of the fertile patient studies was Wilcox 22% (43 of 198), Ellish 17% (20 of 115), Zinaman 13% (15 of 116) and the average rate was 18% (78 of 429). In the infertile population, the average biochemical pregnancy rate was 13.8% (98 of 706). This was composed of a biochemical pregnancy rate of 13.8% in fresh IVF cycles and of 14.4% in FET cycles.

The Ellish study was based on a population of women from the Albany, New York region. The Wilcox study sample was collected in New York City, while the Zinaman articles' subjects hailed from the Washington, DC, USA region. The IVF population in this study was collected in Montreal, Quebec, Canada. All these compared groups were collect in multinational North American cities with large immigrant populations and likely were very similar.

We would also have expected a higher rate of biochemical pregnancy in our population due to mean age being older than the general fertile population (34 versus 29.4 years). It can be hypothesized that the transfer of a selected high quality embryo reduced the chance of biochemical pregnancy. This hypothesis is supported by the findings of the logistic regression analysis which found biochemical pregnancy rates to be lower with blastocyst than with cleavage stage transfer (Table III). Clearly better embryo quality gives lower biochemical pregnancy rates.

The biochemical pregnancy rate in fresh IVF cycles was noted to be similar to the frozen cycles (13.8 versus 14.4%). It could be theorized that the rate of biochemical pregnancies should be increased in fresh cycles versus frozen cycles. Shapiro et al., (2011) reported implantation failure rate of 16% of transfers in the cryopreservation group and 45.3% of transfers in the fresh group. This is thought to be due to alterations in endometrial receptivity. Lessey (1998) demonstrated that decreased beta integrin expression not only lowers implantation rates, but that those embryos which implant have higher miscarriage rates. It is possible that once implantation occurs the biochemical miscarriage rates are unaffected by the changes which are seen in either fresh or frozen IVF cycles (Table I). The embryos transfers in fresh cycles were performed Days 2–5 post-collection. However, the embryos transfers in frozen cycles were mostly blastocyst. It remains unknown how the interaction between cleavage stage embryos in fresh cycles and Day 5 embryos in frozen cycles, with different alterations in endometrium seen in each of these types of cycles, altered these findings. Particularly, in the context of blastocysts seeming to have lower biochemical pregnancy rates than cleavage stage embryos. We did not compare fresh blastocyst to frozen because the policy in our clinic is to transfer blastocysts to most of our patient whenever possible and to freeze only blastocysts. However, this would be an interesting study to perform in the future.

It could be argued that that the biochemical pregnancy rates in the IVF cycles were similar to and not lower than that in the fertile population since one of the fertile population studies had a similar rate to that of the infertile patients (14% versus Zinaman 13%). This however would remain an important finding since it would suggest that the alterations in the endometrium seen in IVF cycles do not have a deleterious effect on the biochemical pregnancy rate. It should never the less be stressed, that the biochemical pregnancy rate of the three studies combined most likely is the most robust estimate of the true rate in the fertile population and was statistically higher than the rate seen in the infertile group.

The difference in detection rates of biochemical pregnancies between the investigated group and the historical control group may have been related to the sensitivity of blood analysis which differs from urinary analysis. From a biological stand point, the recognition of early pregnancy almost certainly missed some losses. In the historical control studies they were able to identify only those pregnancies in which intact hCG was secreted into the urine. This was not the case in the IVF population, who underwent blood analysis for hCG.

It could be hypothesized that in subjects stimulated with higher doses of gonadotrophins, increased alterations of endometrial gene expression occurs altering the likelihood of a biochemical pregnancy. In this study, total gonanadotropin dose failed to be a predictor of biochemical pregnancy (Table III).

A potential weakness of this study is that it is retrospective and is without an in house control group. It clearly would be extremely difficult to recruit an in house control group. Therefore, given the difficulties, the best option to control for the biochemical pregnancy rate was to examine a historical control group. It is unlikely that time differences between the studies occurrences lead to the decreased rate of biochemical pregnancies seen in the infertile patients. We would not expect biochemical pregnancy rates to vary in a 20-year time period without any occurrences of severe stresses, radiation exposure or famine.

In conclusion, these results suggest that there is evidence that IVF may reduce the biochemical pregnancy rate in women under 42 years of age when compared with spontaneous pregnancies. Fresh IVF does not appear to impact the biochemical pregnancy rates when compare with frozen IVF cycles.

Authors’ roles

A.Z. was involved in the planning of the study, IRB proposal authorship and submission, collection of data, analysis of data and manuscript preparation. W.Y.S. was involved in the planning of the study, IRB proposal authorship and submission, collection of data, analysis of data and manuscript preparation. H.M. was involved in the, collection of data, analysis of data and manuscript preparation. M.H.D. was involved in the planning of the study, IRB proposal authorship and submission, collection of data, analysis of data and manuscript preparation, and was the senior author on the study, formatting the initial idea.

Funding

No external funding was either sought or obtained for this study.

Conflict of interest

None declared.

References

Author notes