Revisiting serum β-hCG cut-off levels and pregnancy outcomes using single embryo transfer

Abstract

Purpose

The objective of this study is to identify the pregnancy outcomes based on day-16 β-hCG level assessed with modern assays, in fresh single embryo transfers.

Methods

A retrospective cohort study at a single academic center between 2013 and 2017. A total of 1076 pregnancies were included.

Results

Pregnancies were divided into 10% groupings of 107–108 patients each. The 10 groups did not differ for baseline characteristics. There was no difference on outcomes based on cleavage or blastocyst transfer. At a serum β-hCG level of 103 ± 13 (range 74–135), 50% had a biochemical loss. Biochemical pregnancy losses remained 21% at serum β-hCG range (136–197). It was only once serum β-hCG level reached 199–252 that the probability of a biochemical pregnancy loss was 12%. Interestingly, if a clinical pregnancy is present even at low day-16 serum β-hCG levels, the likelihood of live birth is approximately 50%. This maximizes to 75% when the serum β-hCG level was at least 253 IU/L. The relationship between serum day-16 β-hCG levels and clinical pregnancy or live birth is quite strong with correlation coefficients above 0.8 which accounted for more than 75% of the variability in outcomes in both cases. Receiver operator curves determined that the cut-off for a clinical pregnancy was 190 and for live birth, it was 213 IU/L.

Conclusion

An increase in the serum β-hCG levels at which to expect a reassuring outcome is required based on modern assays, as compared with the old cut-off levels.

Introduction

Human chorionic gonadotropin (hCG) is a glycoprotein secreted during pregnancy by developing trophoblast cells of the placenta. Other types of tissues, including hyperplastic and malignant cells as in choriocarcinoma and other neoplasms, also secrete it [1]. HCG is composed of two glycosylated-distinguished subunits called alpha and beta (β) subunits. The alpha subunit has 92 amino acids, which are identical to the pituitary luteinizing hormone, follicular stimulating hormone, and thyroid-stimulating hormone. The β-subunit of hCG is distinct and has 145 amino acids. It is responsible for the biological activity of the hCG [2, 3].

As pregnancy advances, the trophoblastic tissue increases in volume, leading to the higher serum levels of beta human chorionic gonadotropin (β-hCG). The secreted β-hCG is considered a qualitative marker of the trophoblastic function [4]. As the gestational sac and embryo grow in size, they secrete larger quantities of β-hCG, eventually peaking in the serum at about 8–10 weeks gestational age [5]. Substantial literature suggests that serum β-hCG levels in early pregnancy are associated with the outcomes and are clinically used as such. In a previous study, low serum β-hCG levels measured between 6 and 8 weeks from last menses were found to be associated with an increased risk of pregnancy loss even in the presence of fetal cardiac activity [4]. In another study, serial serum β-hCG measurements were performed from the time of conception and at every 3–4 day intervals in patients with recurrent pregnancy losses. This study demonstrated that peak serum β-hCG levels of at least 88,000 IU/mL were predictive of a successful early pregnancy [6].

The presence of ultrasound evidence of an intra-uterine gestational sac is known as a clinical pregnancy. A positive serum or urinary β-hCG in the absence of the development of ultrasound evidence of pregnancy that returns to zero is known as a biochemical pregnancy loss [7]. Biochemical pregnancy losses usually go undetected because they may occur before the patient even misses her menses. In vitro fertilization (IVF) acts as an excellent model to investigate the role of early serum β-hCG levels in predicting pregnancy outcomes as the embryo age can be precisely dated. β-hCG levels related to biochemical pregnancy loss have become clearer in the era of single embryo transfer. When multiple embryos are transferred, a viable pregnancy can mask a concurrent biochemical pregnancy loss. Therefore, single embryo transfer is crucial to understand the relationship between serum hormonal values at 16 days embryo age (4 weeks gestational age) and pregnancy outcomes.

Many studies have investigated the role of serum β-hCG levels in early pregnancy as a predictor of pregnancy outcomes in IVF cycles. In one study from 2011, serum β-hCG was measured on day 12 after embryo transfer (ET), day 17 embryo age, and levels higher than 80 IU/mL were found to be associated with positive fetal cardiac activity seen on the ultrasound at 6 weeks gestational age [8]. Naredic et al. demonstrated that a serum β-hCG level less than 500 IU/L on day 16 after either fresh or frozen ET was associated with adverse pregnancy outcomes and values more than 500 IU/L were associated with a higher rate of ongoing pregnancy after the first trimester (this equated to day 21 embryo age) [9]. This study included an average of two embryos transferred in both groups but with ultrasound findings of only singleton pregnancies. The incidence of biochemical pregnancy loss was estimated at 15–20% after frozen embryo transfer cycles [10, 11] with multiple embryos returned. In a study including only single embryo transfer [7], this incidence was found to be 14%, and the rates were also equivalent in fresh and frozen single embryo transfers [7].

In general, the lower the level of serum β-hCG is, the lower the chances of an ongoing pregnancy. However, most studies included multiple embryo transfers. The levels were likely affected by concurrent biochemical miscarriages combined with viable pregnancies in some cases. A second confounder would be twin biochemical miscarriages. Few studies investigated this issue using a single embryo transfer.

Another confounder is that the sensitivity of β-hCG assays increases over time, as do all biologic assays. Few studies were conducted using the assays currently available. The traditional cut-off for a viable pregnancy at the time of expected menses was a serum β-hCG level greater than or equal to 100 IU/L, which was established over 30 years ago [12]. Given the changes in assay sensitivity, this value has to be re-investigated.

Therefore, this study aims to identify the serum β-hCG level using the currently used highly sensitivity-automated hCG assays, through which the chance of biochemical pregnancy is the lowest, the level at which the risk of biochemical pregnancy normalizes, and the relationship between β-hCG level and pregnancy outcomes.

Materials and methods

This work is a retrospective cohort study conducted at a single academic fertility center between January 2013 and December 2017. A total of 1076 pregnancies resulting from a single fresh embryo transfer were included after the result of the day 16 (embryo age) β-hCG level. The exclusion criteria included the following: more than one embryo transferred, pregnancies with more than one gestational sac identified, ultrasound or surgically confirmed ectopic pregnancies, and those with the β-hCG level done before or after day 16 embryo age. None of the subjects had diabetes mellitus, cardiovascular disease, took medications in class D or X, or had any medical contra-indications to pregnancy. When the pregnancy test was positive and the β-hCG level was greater than 100 IU/L, the patient was scheduled for a viability ultrasound in 1 to 2 weeks. The ultrasound was repeated every few days until an intra-uterine gestational sac was detected and fetal cardiac activity occurred. If the initial day 16-serum β-hCG level was less than 100 IU/L, the serum test was repeated at 2-day intervals until evidence of the pregnancy outcome and location could be determined.

The blood samples were assayed on the Immulite 2500 (Diagnostic Products Corporation, Los Angeles, CA, USA) for the quantitative measurement of β-hCG. The Immulite uses a solid-phase two-site chemiluminescent immunometric assay with a sensitivity of 1 mIU/mL and a calibrated range to 5000 mIU/mL. The intra- and inter-assay coefficients of variation were less than 7%, respectively. The equipment was checked daily using calibration solutions and operated by a single technician with a Bachelor of Science degree.

Regarding quality control, the laboratory director updated the standard operation procedure regularly after discussing with the laboratory team according to evidence and testing. All the equipment in the laboratory was calibrated regularly depending on the equipment requirements, and the official documents were kept in the laboratory. All reagents, media, and chemicals were recorded as they were received, and the expiration dates were noted. Outdated materials were discarded. The temperature and gases for all incubators were monitored daily. Lighting and temperature in the laboratory were controlled and modified as required. All the important equipment related to gametes/embryos was monitored using an electric monitoring system. HEPA/activated carbon permanganate filters were installed in the ceiling of the IVF laboratory. Air quality was monitored regularly to replace filter systems and to ensure that air quality in the IVF laboratory was appropriate for gametes, embryos, laboratory personnel, and patients. The pH was checked when a new batch of media/culture oil was received. Moreover, the sperm survival test and the parallel culture of the sibling embryos to the blastocyst stage were performed to compare the current with the new media batch to determine whether the new batch of media was good to use. For quality assurance, the laboratory director analyzed the results every 3 months and reported to the clinical director and manager if there was a problem with the results. Otherwise, the laboratory director reported on a yearly basis. The laboratory director also maintains the statistics for all the laboratory personnel. These data included indicators such as the intracytoplasmic sperm injection results (fertilization and pregnancy rates), the survival and pregnancy rates after embryo vitrification/warming, and the pregnancy rates after embryo transfer. This information was then used to develop training programs for specific embryologists and to assess performance over time. In addition, the laboratory director analyzed each clinician’s embryo transfer results and reports to the clinical director regularly. Andrologists participated in the external quality assessment for semen analysis regularly to determine whether our semen analysis was within the range of other clinics. If there was any problem, the laboratory director would ask for a quality assessment meeting with the clinical director and manager to solve the problem. The appropriate documents were kept.

Biochemical pregnancy loss has different definitions in the literature. Sher defined biochemical pregnancy as a positive serum β-hCG level after which the pregnancy fails to progress to the ultrasound confirmation [13]. Some defined it as a transient positive β-hCG of less than 100 mIU/mL, which decreases rapidly before the woman misses her menstrual cycle [14]. De Neubourge et al. considered biochemical pregnancy as an early pregnancy loss before 13 weeks of gestation. After a single good quality embryo transfer, the pregnancy (clinical or biochemical) was diagnosed once there were two increasing values of β-hCG greater than 5 mIU/mL [15]. In our practice, we defined biochemical pregnancy loss according to Sher.

Clinical pregnancy is defined by the International Committee for Monitoring Assisted Reproductive Technology and the World Health Organization as a pregnancy with at least one gestational sac visualized by ultrasonography or the presence of clinical signs of pregnancy. This definition includes ectopic pregnancy. In our practice, clinical pregnancy was defined as clinical pregnancy once one gestational sac was seen by ultrasound and excluded ectopic pregnancy. We defined live birth as a live baby born after 24 weeks of gestation.

Statistical analysis

Statistics were compared using SPSS 23.0 (IBM Corporation, Chicago, USA). Continuous data were evaluated for normalcy using the Kolmogorov–Smirnov test. All continuous data were normally distributed. The continuous data were analyzed using one-way ANOVA. The relationships between data were determined with the Pearson correlation coefficient and the confidence interval for r when appropriate. Categorical data were analyzed using chi-square testing. The receiver operator curves were plotted to determine the cut-off for the pregnancy outcomes. Data are presented as the mean ± standard deviation or percentage. Research Ethics Board approval of the study was obtained (no. 2019-5548).

Results

In total, 1076 pregnancies, as detected by serum β-hCG levels after a single embryo transfer, were divided into 10% (decile) groupings of 107 or 108 patients each. As programmed, the serum β-hCG levels increased across the groups [p < 0.0001, 18 ± 7, 50 ± 13, 103 ± 19, 167 ± 19, 223 ± 15, 287 ± 21, 350 ± 20, 432 ± 26, 547 ± 49, and 1274 ± 1556 at a range of 8–30, 31–73, 74–135, 136–197, 199–252, 253–317, 318–389, 391–478, 479–632, and 633–10,764, respectively]. The 10 groups did not differ in female age (p = 0.07), male age (p = 0.46), number of follicles less than 14 mm, number of follicles greater than or equal to 14 mm (p = 0.10), number of oocytes collected (p = 0.10), number of 2pn (p = 0.43), % fertilized (p = 0.54), number of cleavage (p = 0.43), day of embryo transfer (p = 0.14), number of blastocysts (p = 0.47), and quality of embryo transfer (p = 0.21) (Table 1).

Table 1.

Baseline characteristics of the 10th percentile groupings of serum beta hCG levels

Serum beta hCG level (IU/L)	Group 1 8–30	Group 2 30–73	Group 3 74–135	Group 4 136–197	Group 5 199–252	Group 6 253–317	Group 7 318–389	Group 8 391–478	Group 9 479–632	Group 10 633–10,764	p
Female age	34.8 ± 4.5	35.1 ± 3.4	35.4 ± 4.3	34.9 ± 4.11	35.2 ± 3.8	34.2 ± 4.1	34.6 ± 4.1	34.5 ± 4.8	33.6 ± 4.1	34.2 ± 4.1	0.07
male age	37.9 ± 5.5	37.9 ± 5.6	38.4 ± 6.7	37.8 ± 8.6	38.3 ± 7.2	37.7 ± 6.8	37.3 ± 5.3	37.7 ± 6.5	36.3 ± 5.2	37.1 ± 5.4	0.46
Follicles ≥ 14	5.6 ± 3.2	5.9 ± 3.2	6.2 ± 3.2	6.8 ± 3.9	5.6 ± 2.6	6.4 ± 3.2	6.3 ± 3.6	6.6 ± 4.0	6.7 ± 4.2	6 ± 3.3	0.1
Follicles < 14	8.2 ± 6	8.9 ± 6.9	8.7 ± 6.1	9.1 ± 6.3	8.1 ± 5.8	8.1 ± 5.3	9.9 ± 7.2	8.8 ± 6.9	8.8 ± 5.9	9.3 ± 7.2	0.56
Collected oocytes	10.1 ± 5.8	10.4 ± 5.9	11.3 ± 3.2	11.1 ± 6.3	9.3 ± 4.6	10.1 ± 5.6	11.6 ± 5.9	10.1 ± 5.2	11 ± 5.8	11 ± 6.2	0.1
Viable oocytes	9.9 ± 5.7	10.2 ± 5.9	11.2 ± 5.7	11 ± 6.3	9.1 ± 4.5	10.1 ± 5.6	11.5 ± 5.9	10 ± 5.2	10.8 ± 5.7	10.8 ± 6.2	0.08
MII oocytes	7.6 ± 4.4	7.8 ± 4.4	9 ± 4.7	8.5 ± 5.2	7.3 ± 4	7.6 ± 4.2	9 ± 5	8.4 ± 4.5	8.7 ± 4.7	8.2 ± 5	0.06
Fertilized 2PN	5.4 ± 3.6	5.7 ± 3.6	6.8 ± 3.9	6.3 ± 3.9	5.5 ± 3.5	5.7 ± 3.7	6.6 ± 3.9	6.2 ± 3.7	6.3 ± 3.9	6.4 ± 4.0	0.07
% Fertilized	72 ± 24	75 ± 21	76 ± 20	76 ± 24	76 ± 24	75 ± 22	75 ± 19	75 ± 20	74 ± 23	90 ± 10	0.06
Cleavage	5.3 ± 3.5	5.5 ± 3.2	6.6 ± 3.8	6.2 ± 3.9	5.3 ± 3.2	5.6 ± 3.6	6.4 ± 3.8	6.1 ± 3.6	6.2 ± 3.9	6.3 ± 4.0	0.12

The likelihood of a biochemical pregnancy loss decreased as the serum β-hCG levels increased (97%, 84%, 50%, 21%, 12%, 7%, 8%, 4%, 2%, and 5%, p < 0.0001). Interestingly, at a serum β-hCG level of 103 ± 13 (range 74–135), 50% experienced biochemical loss. Biochemical pregnancy losses remained at 21% at the serum β-hCG range of 136–197. It was only once serum β-hCG level was 199–252 that the probability of a biochemical pregnancy loss was 12%. When the serum β-hCG level reached the range of 391–10,764, the probability of a biochemical pregnancy loss was at its lowest in comparison with the range of 199–389 IU/L (p = 0.003) (Fig. 1a).

Fig. 1.

Chi-square test was done: a The likelihood of a biochemical pregnancy loss decreased as serum β-hCG levels increased (p < 0.0001). b There was no difference in the correlation of biochemical miscarriages rates and β-hCG grouping when groups were divided into blastocysts and cleavage stage embryo transfer cycles. Cleavage r = 0.67, p = 0.0001, 95% CI (0.59–0.74). Blastocyst r = 0.62, p = 0.0001, 95% CI (0.58–0.66)

The association between the β-hCG level and the biochemical pregnancy loss in the blastocyst and cleavage stage embryo transfer cycles was analyzed. No difference was found in the correlation between biochemical miscarriage rates and β-hCG groupings when the groups were divided into the blastocyst and cleavage stage embryo transfer cycles. The cleavage cycle was r = 0.67, p = 0.0001, 95% CI (0.59–0.74), and the blastocyst cycle was r = 0.62, p = 0.0001, 95% CI (0.58–0.66) (Fig. 1b). This can be determined by the overlap in the confidence intervals of the correlation coefficients.

An evaluation of the clinical pregnancy rates and the live birth rates is shown in Fig. 2a. The likelihood of live birth or clinical pregnancy reached the maximum at the serum β-hCG level of 253 IU/L or higher, at which point it plateaued. Moreover, the likelihood of live birth was low at the serum β-hCG level of 135 IU/L or less. This is the level we considered the patient to be safe with older assays. Interestingly, if a clinical pregnancy was present even at low day 16 serum β-hCG levels, the likelihood of live birth was approximately 50% (Fig. 2b). This rate could be maximized 75% when the serum β-hCG level was at least 253 IU/L. Note that the loss rate after clinical pregnancy was at least 20–25% in the best case scenario in this large cohort of IVF patients with a single embryo transfer (Fig. 2b). The relationship between serum day 16 β-hCG levels and clinical pregnancy or live birth was strong, with correlation coefficients above 0.8, which accounted for more than 75% of the variability in outcomes in both cases (Fig. 2a). Even among women with clinical pregnancy, the likelihood of live birth remained tied to the initial serum β-hCG level on day 16, again accounting for 75% of the variability in outcomes that were observed (Fig. 2b).

Fig. 2.

a This figure showed a strong positive correlation between serum β-hCG level and clinical pregnancy rate and LBR, r = 0.8665 (p = 0.001), r = 0.9377 (p = 0.00061). b There is a strong positive correlation between serum β-hCG level and LBR in those with clinical pregnancy. Pearson correlation coefficients r = 0.8628, p = 0.001

To further investigate the role of serum β-hCG levels on day 16 embryo age on the outcomes, we plotted the received operator curves (Fig. 3a, b, and c). We determined the live birth and clinical pregnancy cut-offs using these receiver operator curves and compared them with the cut-offs determined above. The cut-off for a clinical pregnancy was 190 IU/L (sensitivity, 81.2; specificity, 87.1%; positive predictive value (PPV), 94%; and negative predictive value (NPP), 65.2%) and that for live birth was 213 IU/L (sensitivity, 80%; specificity, 71%; PPV, 75.7%; and NPP, 75.7%). The chances for a biochemical pregnancy loss would be high if the serum β-hCG level on day 16 was under 141 IU/L.

Fig. 3.

Receiver operator curve: the area under a ROC curve represents a popular measure of the accuracy of a diagnostic test. For the live birth variable, the area under the curve is equal to 0.828 and for the clinical pregnancy, the area under the curve is equal to 0.91; those values are higher and that indicate better test performance. The area under the curve summarizes the ability of the test to discriminate between patients with true positive results versus patients with false positive test results. The possible values of the area under the curve range from 0.5 (no diagnostic ability) to 1.0, which indicate perfect diagnostic ability. a For the live birth variable, results show that the threshold value to exclude patients with safety is 213 IU/L (sensitivity 80%, specificity 71%, PPV 75.7%, NPP 75.7%). b For the clinical pregnancy variable, the threshold value to exclude patients with safety is 190 IU/L (sensitivity 81.2, specificity 87.1%, PPV 94%, NPP 65.2%). c For the biochemical pregnancy variable, the most appropriate cut-off value is 141 IU/L

Discussion

In this study, fresh embryo transfer cycles were investigated to evaluate the cut-off levels of the serum β-hCG and pregnancy outcomes. Previous studies examined the pregnancy outcome in both fresh and frozen embryo transfer cycles and found no difference in the biochemical, ectopic, and clinical miscarriage rates [7, 16].

Few studies also used a single embryo transfer to eliminate the confounding effect of twin pregnancy associated with multiple embryo transfers on determining the levels of serum β-hCG and its relationship with biochemical pregnancy loss rates. As previously mentioned, Naredic et al., who used a chemiluminescent microparticle enzyme immunometric assay (maker unspecified), demonstrated that a serum β-hCG level less than 500 IU/L at the embryo age of 21 days after both fresh and frozen embryo transfers was associated with adverse pregnancy outcomes. Conversely, serum levels greater than 500 IU/L were associated with higher rates of ongoing pregnancy past the first trimester [9]. This study included an average of two embryos transferred in both groups but with ultrasound findings of only singleton pregnancies. It also involved subjects 5 days more pregnant than those in our study and 5 days after than menses was expected. This 5-day delay was far from when most IVF centers tested the β-hCG levels in their patients. Another retrospective study by Kathiresan et al., which included only fresh day 3 and day 5 embryo transfers between July 2004 and January 2010, found that the day 5 embryo transfer was predictive of a higher serum β-hCG level conducted on day 15 after fertilization. They also found that the serum β-hCG levels of 78 IU/L and 160 IU/L were predictive of ongoing pregnancy after day 3 and day 5 embryo transfers in 96% and 91% of cases, respectively [17]. They used an Immulite hCG assay, but it was not limited to single embryo transfers. In our study, we found no difference in the correlation between the serum β-hCG levels and outcomes in comparing the cleavage and blastocyst embryo transfers. Explaining this difference in the findings when comparing the two studies is difficult. The difference may be due to our larger population of pregnancies of approximately 1100 in comparison with that of 700 in Kathiresan et al.’s study. Another possibility is that culture or laboratory conditions alter the outcomes and the subsequent serum β-hCG level. Further testing is needed to confirm this result.

As indicated in the literature review, other studies used different hCG assays. Some studies did not specify which hCG assay was used, and some combined more than one type of hCG assay. For example, Liu et al. used Access Total β-hCG [6], and Morse et al. used both Abott AxSYM and DPC Immulite assays [5]. In general, different β-hCG assays do not generate the same results on the same specimen.

Some studies used the Immulite 2000 (previous generation) hCG assay, but they were not limited to a single embryo transfer as in Shamonki et al. and Kathiresan at al. [17, 18]. As the currently automated hCG assays are more sensitive to detect different types of hCG molecules, we decided to re-evaluate the hCG levels at which the rate of biochemical pregnancy is the lowest.

In our laboratory, we used the Immulite 2500 assay to detect the serum hCG levels. The Immulite total hCG was compared with the other types of automated total hCG assays, and it was found to be highly sensitive to 8 out of the 9 hCG molecules, including hyperglycosylated hCG, which is the principal hCG molecule in early pregnancy [19]. In addition to the closely monitored serum hCG in assisted reproduction, this assay improved the detection rate of biochemical pregnancy.

In the groups with day 16 serum β-hCG levels above 253 IU/L, the biochemical pregnancy loss rate varied at 2–7% (Fig. 1a). The biochemical miscarriage rate also seemed to increase in the group with the highest β-hCG level in comparison with that in the previous groups with lower levels. However, this is not a statistical difference, as it only represents the expected biological variability.

The main strength of our study is that it included only single embryo transfer cycles, thus eliminating the confounding effect of twin pregnancies associated with multiple embryo transfers. Moreover, only Immulite 2500 was used to measure the serum β-hCG level in all subjects, thus giving consistent results of the serum β-hCG across the groups. Although this is a retrospective study, all data were collected prospectively at the time of pregnancy testing by our head of embryology. This step is a standard at our center for the reporting of IVF and pregnancy outcomes to our national monitoring service. Thus, enrolment bias and loss to follow up were minimized (none were lost).

In conclusion, half of the subjects with a serum β-hCG level of 100 IU/L would have a biochemical pregnancy loss, the level at which the miscarriage rates were believed to be minimized using old assays. The rates of biochemical pregnancy loss were high, with serum β-hCG levels on day 16 of up to 141 IU/L. The risk of biochemical pregnancy loss normalized once serum β-hCG was 191 IU/L at about 14%, the rate expected after IVF as previously determined in another study [3]. The rates reached their lowest level when the serum β-hCG levels were 400 IU/l on day 16 of embryo development. As per counseling, most patients miscarried at serum β-hCG levels of 100 IU/L. The probability of a live birth was reassuring once the day 16 serum β-hCG level was 213 IU/L. Clearly, an increase in the β-hCG levels at which to expect a reassuring outcome was required based on modern assays in comparison with the old cut-off levels based on previous assays. We believe that this result represents new knowledge in the field of reproductive endocrinology and infertility.

Compliance with ethical standards

Research Ethics Board approval of the study was obtained (no. 2019-5548).