Assessing the risk of ectopic pregnancy following euploid frozen embryo transfer: a retrospective cohort study

Abstract

Purpose

To evaluate possible risk factors for an ectopic pregnancy after euploid frozen embryo transfer (FET) at blastocyst stage.

Methods

Retrospective cohort study of women with a positive pregnancy test after euploid frozen embryo transfer at blastocyst stage between April 2017 and June 2023.

Results

An ectopic pregnancy occurred in 21 patients (1.19%) out of 1758 patients. Patient characteristics age, BMI, smoking status, infertility type, cycle/embryo characteristics (type of endometrial preparation and embryo quality), and presence of an isthmocele (yes/no) were not significantly different between the patients with versus without an ectopic pregnancy. The ratio measured to expected hCG levels was significantly lower in patients with ectopic pregnancy when compared to those with clinical pregnancy (0.7 vs 1, P < 0.001). The multinomial regression analysis showed increased odds of an ectopic pregnancy (EP) compared to an ongoing clinical pregnancy with lower observed/expected hCG levels in both single and double embryo transfers (OR, 0.4; 95% CI, 0.07–0.25; P < 0.001 and OR, 0.16; 95% CI, 0.09–0.30; P < 0.001).

Conclusion

The risk of ectopic pregnancy in euploid frozen embryo transfer cycles at the blastocyst stage is comparable to spontaneous conceptions. Despite the huge sample size of euploid FET cycles, due to the very low number of ectopic pregnancies, we were unable to identify a particular risk factor for ectopic implantation. 

Supplementary Information

The online version contains supplementary material available at 10.1007/s10815-025-03560-z.

Introduction

Ectopic pregnancy (EP) results from the implantation of the blastocyst outside the uterine cavity. Different implantation sites have been described, including the fallopian tube, the ovary, the cervix, the cornua, the caesarean scar, and in the abdomen.

Whereas the exact pathophysiology leading to an ectopic implantation of the embryo is not known, the predisposing risk factors suggested are maternal age, smoking, alcohol consumption, tubal disease, history of sexually transmitted disease, intermittent use of clomifen citrate (CC) over an extended period, previous history of EP, and Assisted Reproductive Technology (ART) [1, 2]. Yet, 50% of EP cases still occur in healthy patients without a known predisposing factor [3]. The incidence of EP worldwide ranges between 1 and 2% in spontaneous conceptions [4]. However, it appears that the EP incidence in fresh embryo transfer cycles is higher in comparison to frozen embryo transfers (FET), reaching up to 8.6% [5–7].

An EP is still a life-threatening condition; therefore, identifying probable risk factors in ART treatments is critical. Because aneuploidy and known pathologic tubal factors might both be risk factors for implantation outside the uterine cavity, removing aneuploidy through euploid frozen embryo transfers and cases with pathological tubal factor as well as with an unclear history of a tubal factor allowed us to focus on other possible EP etiologies.

Materials and methods

This retrospective study included patients with a positive serum hCG after euploid FET of blastocyst stage embryos at a tertiary referral Assisted Reproductive Technology center between April 2017 and June 2023. Patients with a known pathological tubal factor as well as patients with an unclear history of a tubal factor were excluded.

Embryo classification, selection, and warming

Blastocysts were categorized into four groups (top, good, fair, and poor) based on the morphological assessment before TE biopsy and on the day of biopsy as per the Gardner and Schoolcraft criteria [8] (Fig. 1).

Fig. 1.

Quality classification of blastocyst based on inner-cell mass, trophectoderm grading, and day of biopsy (“1” = top, “2” = good, “3” = fair, “4–7” = poor)

For the embryo transfer, the selected euploid embryo(s) was/were warmed using the Cryotop method (Kitazato BioPharma Co., Ltd., Shizuoka, Japan) and incubated for 2–4 h to allow blastocoel re-expansion before transfer.

The decision of single embryo transfer (SET) or double embryo transfer (DET) was based on patients’ reproductive and obstetric history as well as patients’ preferences. Despite extensive counseling on the complications associated with DET and subsequent multiple gestations, some patients still opted for DET.

FET protocols

The programmed cycles (PC) or natural cycles (NC) were employed for endometrial preparation at physicians’ and patients’ discretion.

Programmed cycle

Estradiol (E2) valerate 4 mg/day orally was started on cycle day 2 or 3 and increased to 6 mg on day 4 of treatment. When the endometrial lining reached a trilaminar pattern, without a predefined threshold for the thickness [9], after 10–16 days of E2 valerate administration, vaginal micronized progesterone (MVP) was initiated. During the study period, different regimens of luteal phase support (LPS) were used, including cycles in which LPS only comprised of MVP and others in which oral progesterone (dydrogesterone) was added to MVP as clinical standard [10]. Embryo transfer was scheduled 120 h after the start of MVP exposure.

Natural cycles

From cycle day 2/3 onwards, women underwent intermittent ultrasound scans and measurements of the serum luteinizing hormone (LH), E2, and progesterone (P4) levels throughout the menstrual cycle to determine the timing of ovulation. The pre-ovulatory LH surge was considered to have begun when the concentration rose by 180% above the most recent serum value and continued to rise thereafter [11]. Ovulation was confirmed when P4 concentrations of 1.0 ng/ml [12] and above were detected, designated as day 0. Blastocyst transfer was performed approximately 120 h after ovulation was identified.

Embryo transfer procedure

Embryo transfers were performed in the lithotomy position under abdominal ultrasound guidance with a full bladder. Patients were allowed to ambulate directly afterwards. Our internal audits suggest similar live birth per embryo transfer per physician in our centers.

In both endometrial preparation approaches, hCG testing was performed 10 days after the embryo transfer, and a level of > 15 IU was regarded as a positive result. When patients achieved a pregnancy, LPS was continued in both endometrial preparation approaches either until 7 weeks (NC) or until 12 weeks (PC) of pregnancy.

Outcome definitions

A clinical pregnancy (CP) was diagnosed when an intra-uterine gestational sac (GS) was seen on ultrasound. An EP was diagnosed with the sonographic finding of a GS outside the uterine cavity, the histopathologic documentation of products of conception collected outside the uterine cavity during surgery and/or based on clinical suspicion in the case of sub-optimal increments of hCG values with absent sonographic evidence of a GS.

Statistical analysis

Association between patient and embryo characteristics and outcomes was assessed with multinomial regression analyses, and results are reported as odds ratios with 95% confidence intervals. P levels below 0.05 were considered statistically significant. Association of hCG levels with measurement days was modeled using generalized additive models for location, scale, and shape. hCG levels were shown to have a best fit for Box-Cox Power Exponential distribution family. After estimating the expected mean hCG levels for each measurement day, observed/expected hCG ratios were calculated.

Results

A total of 1758 patients with a positive hCG test after the transfer of euploid embryos at blastocyst stage were included, of which 642 (36.5%) were DETs and 1116 (63.4%) SETs. Out of 1758 pregnancies, 21 were identified to be of ectopic origin (1.196%), 195 were biochemical (11.09%), 267 were clinical miscarriages (15.18%), and the remaining 72.52% were clinical pregnancies. Patient characteristics age, BMI, smoking status, infertility type, cycle/embryo characteristics (type of endometrial preparation and embryo quality), and presence of an isthmocele (yes/no) were not significantly different between the patients with versus without an EP (Table 1).

Table 1.

Patient and embryo characteristics in ectopic and biochemical pregnancy, clinical miscarriage, and clinical pregnancy groups

Variables		Ectopic pregnancy	Biochemical pregnancy	Clinical miscarriage	Clinical pregnancy	P*	P**	P†
Female age (years)	Median (IQR)	32.0 (28.0 to 35.0)	32.0 (28.0 to 36.0)	33.0 (29.0 to 37.0)	33.0 (29.0 to 37.0)	0.838	0.542	0.403
BMI (kg/m2)	Median (IQR)	25.5 (23.7 to 29.8)	26.7 (23.3 to 29.9)	27.6 (24.5 to 31.5)	26.2 (23.0 to 30.1)	0.438	0.049	0.466
Transfer protocol	Natural (n/%)	5 (23.8)	60 (30.8)	69 (25.8)	510 (40.0)	0.682	0.999	0.201
	Programmed (n/%)	16 (76.2)	135 (69.2)	198 (74.2)	765 (60.0)
Infertility type	Primary (n/%)	11 (52.4)	98 (50.3)	140 (52.4)	612 (48.0)	0.999	0.999	0.858
	Secondary (n/%)	10 (47.6)	97 (49.7)	127 (47.6)	663 (52.0)
Number of transferred embryos	Single (n/%)	12 (57.1)	140 (71.8)	164 (61.4)	800 (62.7)	0. 252	0.877	0.765
	Double (n/%)	9 (42.9)	55 (28.2)	103 (38.6)	475 (37.3)
Isthmocele	No (n/%)	19 (90.5)	175 (89.7)	226 (84.6)	1130 (88.6)	0.999	0.686	0.999
	Yes (n/%)	2 (9.5)	20 (10.3)	41 (15.4)	145 (11.4)
Smoking	No (n/%)	21 (100.0)	190 (97.4)	252 (94.4)	1217 (95.5)	0.999	0.545	0.640
	Yes (n/%)	0 (0.0)	5 (2.6)	15 (5.6)	58 (4.5)
Embryo quality	Top (n/%)	7 (33.3)	27 (13.8)	51 (19.1)	306 (24.0)	0.087	0.381	0.696
	Good (n/%)	11 (52.4)	118 (60.5)	161 (60.3)	749 (58.7)
	Fair (n/%)	2 (9.5)	17 (8.7)	23 (8.6)	100 (7.8)
	Poor (n/%)	1 (4.8)	33 (16.9)	32 (12.0)	120 (9.4)
Measured/Expected hCG	Median (IQR)	0.7 (0.6 to 0.7)	0.6 (0.5 to 0.8)	0.9 (0.8 to 1.0)	1.0 (0.9 to 1.1)	0.550	< 0.001	< 0.001

^*Ectopic pregnancy vs biochemical pregnancy; **ectopic pregnancy vs miscarriage; †Ectopic pregnancy vs clinical pregnancy

The ratio of measured to expected hCG levels was significantly lower in patients with an EP when compared to those with a CP (median, 0.6 vs 1.0, P < 0.001). Moreover, measured to expected hCG levels were lower in biochemical pregnancies and miscarriages compared to ongoing clinical pregnancies. The measure observed to expected hCG levels of EP closely resembled that of biochemical pregnancies (median, 0.7 vs 0.6, P = 0.55) and were significantly lower compared to miscarriages (median, 0.7 vs 0.9, P < 0.001) (Table 1). Figure 2 depicts the measured to expected hCG levels in biochemical pregnancies and miscarriages compared to ongoing clinical pregnancies for the SET cycles.

Fig. 2.

hCG levels according to the day of testing and measured/expected hCG ratios stratified by outcome for single euploid frozen embryo transfers. BP, biochemical pregnancy; CP, clinical pregnancy

The multinomial regression analysis showed a slight but statistically significant increased odds ratio of an EP compared to an ongoing clinical pregnancy with a lower ratio of observed/expected hCG levels in both single and double embryo transfers (OR, 0.14; 95% CI, 0.07–0.25 and OR, 0.16; 95% CI, 0.09–0.30; both P < 0.001, respectively (Table 2 and Table S1)). None of the patients with an EP was smoking; therefore, the factor “smoking” could not be assessed as a potential risk factor for an EP. The results of the multinomial regression analysis for other outcomes (biochemical and miscarriage) are presented in Table 2.

Table 2.

Multinomial regression analysis looking at factors associated with biochemical pregnancy (BP), ectopic pregnancy (EP) and clinical pregnancy (CP), SET (single embryo transfer) only

Predictors	Odds ratios	CI	p	Response
Clinical vs biochemical
Transfer protocol
• Natural	Reference
• Programmed	1.66	0.99–2.78	0.055	CP vs BP
Embryo quality
• Top	Reference
• Good	1.58	0.80–3.13	0.191	CP vs BP
• Fair	0.98	0.33–2.88	0.974	CP vs BP
• Poor	1.14	0.47–2.74	0.773	CP vs BP
Body mass index	0.97	0.92–1.02	0.251	CP vs BP
Smoking	0.34	0.07–1.65	0.179	CP vs BP
Observed/Expected hCG	0.12	0.09–0.16	< 0.001	CP vs BP
Clinical vs miscarriage
Transfer protocol
• Natural	Reference
• Programmed	1.72	1.18–2.51	0.005	CP vs miscarriage
Embryo quality
• Top	Reference
• Good	1.33	0.83–2.12	0.233	CP vs miscarriage
• Fair	1.33	0.63–2.81	0.452	CP vs miscarriage
• Poor	1.17	0.59–2.32	0.654	CP vs miscarriage
Body mass index	1.03	1.00–1.07	0.072	CP vs miscarriage
Smoking	1.51	0.70–3.25	0.296	CP vs miscarriage
Observed/Expected hCG	0.42	0.33–0.52	< 0.001	CP vs miscarriage
Clinical vs ectopic
Transfer protocol
• Natural	Reference
• Programmed	3.83	0.80–18.39	0.094	CP vs ectopic
Embryo quality
• Top	Reference
• Good	0.38	0.10–1.44	0.155	CP vs ectopic
• Fair	0.40	0.04–3.91	0.429	CP vs ectopic
• Poor	0.20	0.02–1.89	0.159	CP vs ectopic
Body mass index	0.94	0.82–1.08	0.368	CP vs ectopic
Smoking	NE	NE	-	CP vs ectopic
Observed/expected hCG	0.14	0.07–0.25	< 0.001	CP vs ectopic

CI, confidence interval

Discussion

The data presented here demonstrated that there is no increase in the risk of an EP after euploid FET at blastocyst stage compared to the EP risk in the general population, which has been reported to be between 1 and 2% [13]; furthermore, it appears to be lower compared to other data from IVF population, with a described incidence of 1.4–5.4% [14]. In ART, irrespective of the patients’ age, fresh embryo transfers and multiple embryo transfers in the setting of tubal pathology have been identified as risk factors for an EP during IVF therapy [15–17]. Based on our analysis, euploid FET cycles do not increase the risk of an EP.

The impact of the endometrial preparation approach on the EP rate is discussed controversially. While in our data the PC endometrial approach resulted in more EPs than the NC approach (1.43% vs 0.7%), the difference was not statistically significant. Zhang et al. found a significant difference between the approaches with higher EP rates in PC compared to NC (3.31% vs 1.46%, respectively; P = 0.045) [18]. However, the overall EP rate in the study of Zhang et al. was 2.39%, which is higher compared to our data and to natural conceptions and might be a result of day 3 embryo (cleavage stage) transfers as well as the multiple embryo transfer strategy with an average of 2.45/2.52 embryos transferred in the NC and the PC approach and without preimplantation genetic testing for aneuploidy. Similar to our results, Atkinson et al., who compared four FET protocols of non-genetically tested embryos, performed as ovulation induction cycle with letrozole, ovulation induction cycle with gonadotropins, a NC and a PC, did not demonstrate any differences with EP rates being 1%, 2.2%, 0.8%, and 1% respectively [19]. Jwa et al. conducted a register-based analysis of 153,354 clinical pregnancies with an EP rate of 0.52% (792 cases) and found no difference between PC and NC preparation approaches but described an increased risk after a modified natural cycle using clomiphene citrate [20]. Further research will have to confirm the influence of the endometrial preparation approach on the risk of an EP and whether hypothetically differences in the hormonal levels between PC and NC or the administration of exogenous hormones could play a role by affecting endometrial receptivity or tubal motility.

It is well known that embryo quality has a significant impact on the ART outcome [21]; therefore, the quality of euploid embryos might also have an influence on the implantation site, as an intricate communication between embryo and endometrium for correct implantation was described previously [22]. In our presented data, we did not observe a significant difference in the embryo quality between patients with versus without an EP. These findings are in line with Zhang et al. [23] who compared the clinical outcomes according to the euploid embryo qualities divided into AA, AB, BA, BB, and any C. In a registry-based study of 15,006 clinical pregnancies (fresh embryo transfer, n = 8952; FET, n = 6054) after the transfer of non-tested embryos, Anzhel et al. reported a protective effect of transferring top quality embryos towards an EP in both fresh and frozen embryo transfer cycles, with an average EP rate of 2% [24]. Furthermore, the developmental stage of the embryo at the time of the embryo transfer appears to have an impact on the risk of an EP in addition to the embryo’s quality. According to Huang et al. [25], transferring embryos in the cleavage stage may raise the likelihood of an EP. We are unable to add any data regarding the embryo developmental stage due to exclusively undertaking blastocyst transfers in our clinic. Inferior embryo quality is often associated with chromosomal aneuploidy [26]; therefore, the higher EP rate of embryos with lower quality might be due to their chromosomal abnormality, leading to the implantation outside the endometrial cavity.

Whereas previous publications suggested that aneuploidy of the embryo is a risk factor for aberrant implantation [27–30], multiple other authors [31–34], including the more recent publication from Ruderman et al. [35], could not confirm these findings. While Coste et al. [31] used also the karyotype to detect aneuploidies of products of conception, Ruderman et al. [35] as well as Furuya et al. [34] resorted to the more advanced single-nucleotide polymorphism microarray as a method of testing. The authors in the latter study, however, based their conclusions on the results of eight tested cases only [35]. Moreover, the platform used could not detect duplications, deletions, balanced tetraploidy, and/or maternal triploidy and mosaicism, which may limit the reliability of the reported findings. If, hypothetically, aneuploid embryos are more likely to implant outside the cavity, increasing accuracy of genetic platforms might minimize the prevalence of EP following euploid FET in the future. Our extensive review of the literature revealed that so far only a limited number of studies focussed on the prevalence of aneuploidy in ectopic pregnancies. The studies are summarized in Table 3 and demonstrate a wide range of aneuploidy rates in the different studies. Further studies into the potential causative involvement of embryo ploidy in the genesis of ectopic pregnancies and the application of unified genetic testing panels to identify aneuploidies are necessary due to the paucity of existing data.

Table 3.

Summary of studies reporting aneuploidy rates in ectopic pregnancies

Study (author(s)/year)	Method	Sample size	Aneuploidy rate
Busch DH, Benirschke K (1974) [39]	Karyotyping	25 EPs	16%
Poland et al. (1976) [28]	Karyotyping	16 EPs	31.30%
Elias S et al. (1981) [40]	Karyotyping	23 EPs	17% (mostly in sac-only samples)
Aine R et al. (1990) [29]	Flow cytometry	42 EPs	~ 33%
Karikoski R et al. (1993) [30]	Flow cytometry	42 EPs	33%
Toikkanen S et al. (1993) [41]	Flow cytometry	55 EPs vs 92 miscarriages	24% vs 8%
Cohen D et al. (1993) [42]	Rapid karyotyping	60 EPs	78%
Han SS, Yang YH (1995) [43]	Karyotyping	16 EPs	12.50%
Goddijn M et al. (1996) [32]	Karyotyping	30 EPs	4.50%
Erel CT et al. (1996) [44]	Flow cytometry	13 EPs	38.50%
Block WA Jr et al. (1998) [45]	Karyotyping	22 EPs	14%
Coste J et al. (2000) [31]	Karyotyping	62 EPs	4.80%
Goddijn M et al. (2005) [33]	QF-PCR	70 EPs	3.70%
McKenzie LJ et al. (2005) [27]	Comparative genomic hybridization	15 EPs (MTX failures)	27%
Furuya M et al. (2017) [34]	SNP microarray	88 EPs	3.40%
Ruderman RS et al. (2020) [35]	SNP microarray	8 EPs	0%

EP, ectopic pregnancy; MTX, methotrexate; SNP, single-nucleotide polymorphism

Measurement of serum hCG presents the gold standard of pregnancy diagnostics, and among clinicians, it is well known that lower than expected hCG levels might point towards an abnormal implantation or a biochemical/non-evolutionary pregnancy. Our findings of a statistically significantly lower ratio of observed over expected hCG values in patients with an EP compared to patients with a clinical pregnancy are consistent with previous publications concerning the significance of the initially measured HCG values and subsequent values [36]. The lower hCG values are caused by the ectopic implantation of the embryo due to its poor trophoblastic invasion and a reduced capacity of secreting hCG. As a practical point, monitoring of serial hCG values is of utmost importance to document an adequate rise of hCG and to rule out EP in case of lower measured than expected hCG values.

The exact mechanism of EP is still unknown and correct embryo implantation requires intricate communication between the euploid blastocyst and the implantation site to ensure an eutopic pregnancy. Recent data suggests that apart from the blastocyst, part of this control is based on an immunomodulatory response involving inflammatory cytokines and chemokines [37]. Studies suggest that proinflammatory factors such as IL-1 and IL-6 play a pivotal role in mediating a proper implantation of the blastocyst within the endometrium cavity [38]. It seems that embryo implantation necessitates a certain level of inflammatory reaction as the site of nidation within the endometrium. In case such an inflammatory response appears in ectopic sites, this pro-implantation environment might result in an EP. Furthermore, it has been suggested that these inflammatory processes also hamper proper embryo transport through the fallopian tubes, hence hindering the blastocyst from reaching the endometrial cavity [37].

In conclusion, compared to spontaneous conceptions, it seems that transferring a euploid embryo at the blastocyst stage in a FET cycle does not increase the likelihood of an EP. Compared to published data from IVF populations, the risk of this event is low, and it appears that transferring a euploid embryo lowers the risk. However, even after a single euploid embryo transfer at the blastocyst stage, hormonal and ultrasonographic monitoring is necessary to ensure the correct embryo implantation because of the possibly life-threatening condition of an ectopic pregnancy.

Despite the huge sample size of euploid FET cycles, we were unable to identify a particular risk factor for ectopic implantation. The retrospective study methodology can be viewed as a limitation; however, it is important to note that the here presented data appear to be the largest sample size of euploid FET cycles, investigating the prevalence of ectopic pregnancies. Besides the sample size, the inclusion of patients’, cycle, and embryo characteristics and exclusion of patients with tubal factors are strengths of our study and allow for detailed evaluation of potential risk factors. Hence, because an EP is a rare event, more euploid FET data will be required for a more accurate evaluation of potential risk variables.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contribution

AS: conceptualization, data curation, writing — original draft. BL: conceptualization, supervision, writing — review and editing of the manuscript. EK: data curation, formal analysis, review of the manuscript. RDG: data curation, review of the manuscript. IK: data curation, review of the manuscript. AB: data curation, review of the manuscript. HF: conceptualization, supervision, writing — review and editing of the manuscript.

Data availability

My manuscript has associated data in a data repository.

Declarations

Ethical approval

The study was approved by the research ethical committee of ART Fertility Clinics, UAE (REFA 101–2306-ABU-011-AS).

Competing interests

The authors declare no competing interests.