https://orcid.org/0000-0002-7941-8340
Abstract
What are the plasma concentrations of dydrogesterone (DYD) and its metabolite, 20α-dihydrodydrogesterone (DHD), measured on day of embryo transfer (ET) in programmed anovulatory frozen embryo transfer (FET) cycles using 10 mg per os ter-in-die (tid) oral DYD, and what is the association of DYD and DHD levels with ongoing pregnancy rate?
DYD and DHD plasma levels reach steady state by Day 3 of intake, are strongly correlated and vary considerably between and within individual subjects, women in the lowest quarter of DYD or DHD levels on day of FET have a reduced chance of an ongoing pregnancy.
DYD is an oral, systemic alternative to vaginal progesterone for luteal phase support. The DYD and DHD level necessary to sustain implantation, when no endogenous progesterone is present, remains unknown. While DYD is widely used in fresh IVF cycles, circulating concentrations of DYD and DHD and inter- and intraindividual variation of plasma levels versus successful treatment have never been explored as measurement of DYD and DHD is currently only feasible by high-sensitivity chromatographic techniques such as liquid chromatography/tandem mass spectroscopy (LC-MS/MS).
Prospective, clinical cohort study (May 2018–November 2020) (NCT03507673); university IVF-center; women (n = 217) undergoing a programmed FET cycle with 2 mg oral estradiol (tid) and, for luteal support, 10 mg oral DYD (tid); main inclusion criteria: absence of ovulatory follicle and low serum progesterone on Days 12–15 of estradiol intake; serum and plasma samples were taken on day of FET and stored at −80°C for later analysis by LC-MS/MS; in 56 patients, two or more FET cycles in the same protocol were performed.
Women undergoing FET on Day 2 or Day 3 (D2, D3, cleavage) or Day 5 (D5, blastocyst) of embryonic development had blood sampling on the 3rd, 4th or 6th day of 10 mg (tid) DYD oral intake, respectively. The patient population was stratified by DYD and DHD plasma levels by percentiles (≤25th versus >25th) separately by day of ET. Ongoing pregnancy rates (a viable pregnancy at >10th gestational week) were compared between ≤25th percentile versus >25th percentile for DYD and DHD levels (adjusted for day of ET). Known predictors of outcome were screened for their effects in addition to DYD, while DYD was considered as log-concentration or dichotomized at the lower quartile. Repeated cycles were analyzed assuming some correlation between them for a given individual, namely by generalized estimating equations for prediction and generalized mixed models for an estimate of the variance component.
After exclusion of patients with ‘escape ovulation’ (n = 14, 6%), detected by the presence of progesterone in serum on day of ET, and patients with no results from LC-MS/MS analysis (n = 5), n = 41 observations for cleavage stage ETs and n = 157 for blastocyst transfers were analyzed. Median (quartiles) of plasma levels of DYD and DHD were 1.36 ng/ml (0.738 to 2.17 ng/ml) and 34.0 ng/ml (19.85 to 51.65 ng/ml) on Day 2 or 3 and 1.04 ng/ml (0.707 to 1.62 ng/ml) and 30.0 ng/ml (20.8 to 43.3 ng/ml) on Day 5, respectively, suggesting that steady-state is reached already on Day 3 of intake. DHD plasma levels very weakly associated with body weight and BMI (R2 < 0.05), DYD levels with body weight, but not BMI. Levels of DYD and DHD were strongly correlated (correlation coefficients 0.936 for D2/3 and 0.892 for D5, respectively). The 25th percentile of DYD and DHD levels were 0.71 ng/ml and 20.675 ng/ml on day of ET. The ongoing pregnancy rate was significantly reduced in patients in the lower quarter of DYD or DHD levels: ≤25th percentile DYD or DHD 3/49 (6%) and 4/49 (8%) versus >25th percentile DYD or DHD 42/149 (28%) and 41/149 (27%) (unadjusted difference −22% (CI: −31% to −10%) and −19% (CI: −29% to −7%), adjusted difference −22%, 95% CI: −32 to −12, P < 0.0001).
Some inter- and intraindividual variations in DYD levels could be attributed to differences in time between last 10 mg DYD intake and blood sampling, as well as concomitant food intake, neither of which were registered in this study. Ninety percent of subjects were European-Caucasian and DYD/DHD blood concentrations should be replicated in other and larger populations.
Daily 10 mg DYD (tid) in an artificial FET cycle is potentially a suboptimal dose for a proportion of the population. Measurement of DYD or DHD levels could be used interchangeably for future studies. The pharmacokinetics of oral DYD and associated reproductive pharmacodynamics need further study.
The trial was financed by university funds, except for the cost for plasma and serum sample handling, storage and shipment, as well as the liquid chromatography–mass spectrometry (LC-MS/MS) analysis of DYD, DHD and progesterone, which was financially supported by Abbott Products Operations AG (Allschwil, Switzerland). Abbott Products Operations AG had no influence on the study protocol, study conduct, data analysis or data interpretation. K.N. has received honoraria and/or non-financial support (e.g. travel cost compensation) from Ferring, Gedeon-Richter, Merck and MSD. A.M. has no competing interests. R.V. has no competing interests. M.D. has received honoraria and/or non-financial support from Ferring and Merck. A.S.-M. has no competing interests. T.K.E. has received honoraria and/or non-financial support from Roche, Novartis, Pfizer, Aristo Pharma, Merck. G.G. has received honoraria and/or non-financial support (e.g. travel cost compensation) from Abbott, Ferring, Gedeon Richter, Guerbet, Merck, Organon, MSD, ObsEva, PregLem, ReprodWissen GmbH, Vifor and Cooper.
ClinicalTrials.gov NCT03507673.
Introduction
The transfer of a frozen embryo (FET) to the uterus can be conducted in a so-called programmed or ‘artificial’ cycle in which spontaneous ovulation is suppressed by exogenous sex-steroid administration. Since follicular development and a corpus luteum are absent in such programmed (artificial) cycles, the establishment of endometrial receptivity and the support and maintenance of early pregnancy completely rely upon the orchestrated exogenous administration of sex steroids (Di Renzo et al., 2016). Different progestogens are used in programmed cycles, such as micronized vaginal progesterone (MVP) (Ghobara et al., 2017), injectable progesterone (Lockwood et al., 2014; Turkgeldi et al., 2020) or oral dydrogesterone (DYD) (Griesinger et al. 2019; Vuong et al., 2021). MVP is known to induce a strong local progestogenic effect on the uterus and endometrium; however, systemic progestogenic exposure may be low in some patients. In women with low serum progesterone levels on day of embryo transfer (ET) after vaginal progesterone administration within an artificial FET cycle, a considerable reduction in ongoing pregnancy rate has been observed (Yovich et al., 2015; Labarta et al., 2017; Alsbjerg et al., 2018; CeDrin-Durnerin et al., 2019; Gaggiotti-Marre et al., 2019; Labarta et al., 2021). Since this phenomenon was detected even despite the administration of a dose as high as 800 mg MVP daily (Labarta et al., 2017; 2021), the sole vaginal route for progesterone administration has come under scrutiny in the context of programmed FET. These observations, taken together with further evidence of suboptimal treatment outcome in FET and fresh cycles when using MVP (Devine et al., 2018; Griesinger et al., 2020), have renewed the interest in FET cycle regimen employing a systemic administration of a progestogen (Ghobara et al., 2017; Vuong et al., 2021).
Oral DYD has high bioavailability, shows little first pass effect, has no estrogenic activity (van Amsterdam et al., 1980; Kuhl and Wiegratz, 2021) and is an alternative to vaginal progesterone for luteal phase support (LPS). DYD has recently attained market approval in numerous countries worldwide for LPS in fresh IVF cycles supported by a large phase III trial program (Tournaye et al., 2017; Griesinger et al., 2018). An individual participant data meta-analysis recently reported a significant increase in the likelihood for a live birth for DYD versus MVP use in fresh IVF cycles (Griesinger et al., 2020). In the context of programmed FET, oral DYD has been tested against MVP in a number of small preliminary randomized clinical trials (RCT) (Salehpour et al., 2013; Zargar et al., 2016; Zarei et al., 2017; Atzmon et al., 2020; Hossein Rashidi et al., 2020; Ozer et al., 2020), however, no clear picture on the efficacy of different regimen of DYD versus different regimen of MVP has emerged from these trials.
An anovulatory, programmed FET cycle provides the ideal environment to study the efficacy of a progestogen for inducing receptivity and supporting pregnancy as the sole progestogenic effect of the administered drug becomes apparent without distortion by endogenous progesterone effects originating from one or several corpora lutea. Measurement of DYD and its active metabolite 20α-dihydrodydrogesterone (DHD) is currently only feasible by liquid chromatography–tandem mass spectrometry (LC-MS/MS). This has impeded studying the association of DYD and DHD plasma levels with clinical outcome. The present study therefore investigated DYD and DHD plasma levels achieved by an oral DYD dose frequently used in the context of LPS, as well as the inter- and intraindividual variation of plasma levels and their association with treatment outcome in artificial FET cycles.
Materials and methods
Aims, design and setting
This study was a prospective, single-center cohort study conducted at a university affiliated center for reproductive medicine between May 2018 and November 2020. Institutional review board approval was granted (Ethical Review Board of the University of Luebeck, Germany, reference number 18-005) and all patients provided written informed consent. The present cohort of patients receiving oral DYD in a programmed FET comes from a larger observational cohort study on patients undergoing FET in various protocols (NCT03507673). The study was conducted to describe DYD and DHD plasma levels at a conventionally used dosage of 10 mg (tid) as well as the inter- and intraindividual variation of these levels and how the observed variation may relate to outcome, in close analogy to a previously published artificial FET study utilizing MVP (Labarta et al., 2017).
Study population and outcomes
The study cohort consists of female infertile patients, aged 18–45 years, undergoing a FET cycle following IVF or ICSI who exclusively had DYD 10 mg per os ter-in-die (tid) as a progestogenic drug for endometrial transformation and support of early pregnancy. Patient inclusion was performed on days 13–15 of oral estradiol intake in a programmed cycle. Exclusion criteria were evidence of ovulation on ultrasound prior to ET (defined as the presence of a follicle ≥14 mm and/or progesterone (P4) ≥1.0 ng/ml on endometrial preparation days 13–15 or day of ET or on day of first serum hCG measurement, respectively). Patients with malformations of the uterus or endometrial abnormalities (on ultrasound or diagnosed by previous hysteroscopy) were also excluded. The ongoing pregnancy rate at 10 weeks of gestation was chosen as primary endpoint. The proportion of patients having a positive pregnancy test, a clinical, an ongoing pregnancy, a miscarriage, a preclinical pregnancy loss and a pictorial blood assessment chart score in the luteal phase were secondary outcomes. All FET cycles were performed after vitrification/warming using an open, manual system (Kitazato vitrification kit VT601, Gynemed, Lensahn, Germany).
Treatment, blood sampling and early pregnancy monitoring
For endometrial proliferation, all patients received oral estradiol valerate (E2) 2 mg (tid) starting from Day 1 of the menstrual cycle. On Days 13–15 of continuous E2 administration, appropriate endometrial preparation and the absence of a pre-ovulatory follicle was confirmed by ultrasound. Additionally, anovulation was affirmed by serum progesterone determination in the anovulatory range (<1.0 ng/ml). Subsequently, patients started intake of oral DYD 10 mg tid and were scheduled for FET on Day 3, 4 or Day 6 of DYD intake in case of transfer of a Day 2 or Day 3 cleavage stage embryo or a Day 5 blastocyst, respectively. Late in the morning of the day of FET, plasma samples were drawn and frozen at −80°C for later analysis of DYD and DHD concentrations. From day of ET onwards, patients were assigned to apply transdermal 2.5 mg E2 gel daily (Gynokadin®, Besins Healthcare Germany GmbH, Berlin, Germany) in addition to oral E2 2 mg (tid) and oral DYD 10 mg (tid). Serum hCG to confirm pregnancy was measured at the earliest nine days and at the latest 14 days after FET. All women achieving a pregnancy continued oral E2 and oral DYD administration up to at least the 8th gestational week (GW). A transvaginal sonography was performed to confirm presence of a gestational sac and pregnancy viability 30–36 days after ET. A clinical pregnancy was defined as a positive heartbeat of the embryo on ultrasound in the seventh GW. A miscarriage was defined as a pregnancy achieving serum hCG levels of at least 1000 IU/l but not progressing further as defined by clinical judgment (e.g. a complete abortion, a missed abortion with an empty fetal sac or with an embryonic structure without heartbeat (8th GW) and/or inadequate further increase in hCG levels) at a later stage. Pregnancies with serum hCG levels above lower reference limit but not exceeding hCG levels >1000 IU/l were defined as preclinical pregnancy loss. The course of pregnancy after 10 weeks of gestation, the occurrence of live birth and child health were monitored by phone calls.
DYD, DHD and progesterone measurement
DYD, DHD and progesterone concentrations were determined in human EDTA plasma using solid supported liquid/liquid extraction 96-well plates, elution with heptane/t-butyl methyl ether and after evaporation and reconstitution in HPLC mobile phase, injection into an LC-MS/MS system. Dydrogesterone-D5, DHD-D6 and progesterone-D9 were used as internal standards. The LC-MS/MS method has been validated in compliance with European Medicines Agency (EMA) guidelines and the validation procedure incorporated a five-batches validation on different days to obtain selectivity, sensitivity, stability and intra- and inter-day accuracy and precision in the concentration range (Lower Limit of Quantification (LLOQ)–Upper Limit of Quantification (ULOQ)) of 0.05–20 ng/ml for DYD and progesterone and 0.5–200 ng/ml for DHD. The method was found to be selective, accurate and precise and all analytes were stable under sample processing conditions. The overall average inter-day precision (% coefficient of variation (CV)) of the low, mid and high quality control (QC) sample levels was 4.3, 3.7 and 2.3% and the mean absolute accuracy, expressed as absolute bias, was 1.1, 1.3 and 3.6% for DYD, DHD and progesterone, respectively. The measurements were performed by Nuvisan GmbH (Neu-Ulm, Germany).
Estradiol measurement
Estradiol concentrations were determined in human EDTA plasma using solid supported liquid/liquid extraction 96-well plates, elution with dichloromethane and derivatization with dansyl chloride. After clean-up and reconstitution in HPLC mobile phase, samples were injected into an LC-MS/MS system. Estradiol-D4 was used as internal standard. The LC-MS/MS method has been validated in compliance with EMA guidelines and the validation procedure incorporated a five-batches validation on different days to obtain selectivity, sensitivity, stability and intra- and inter-day accuracy and precision in the concentration range (LLOQ–ULOQ) of 5–500 pg/ml. The method was found to be selective, accurate and precise and all analytes were stable under sample processing conditions. The overall average inter-day precision (%CV) of the low, mid and high QC sample levels was 3.5% and the mean absolute accuracy, expressed as absolute bias, was 0.7%. The measurements were performed by Nuvisan GmbH (Neu-Ulm, Germany).
Sample size
The sample size considerations, performed ex-ante, assumed that DYD and DHD levels would considerably vary and that low levels would be associated with a lower chance of pregnancy achievement. Based on previous observational data on artificial FET cycles (Labarta et al., 2017), it was assumed that the patients in the lower quarter of DYD levels would have a strongly reduced chance of an ongoing pregnancy. Group sample sizes of 44 in the lower quarter (≤25th percentile for DYD levels) and 132 above the 25th percentile achieve 80% power at a significance levels of 0.05 to detect an odds ratio between the group proportions estimated ongoing pregnancy rate of 30% in patients >25th percentile and 10% in those below (two-sided continuity corrected χ2 test).
Statistical analysis
Descriptive statistics (mean, SD, CI or median, quartiles and range or absolute numbers and relative proportions) were utilized for characterization of patient demographics. For all analyses, only the first cycle was analyzed, except for repetitive measurements for comparison of intraindividual DYD/DHD levels between different treatment cycles. Empirical distribution functions (Supplementary Fig. S1) confirmed the anticipated lognormal distributions, so that concentration data were log-transformed before analysis and before entering regression analysis. The patient sample of this study was divided into four groups according to the 25th percentile (or lower quartile); 50th percentile (median) and 75th percentile (upper quartile) of DYD and DHD concentrations, respectively. Live birth rates by DYD and DHD quartiles were analyzed adjusted by time point of ET with fixed-effects, Mantel-Haenszel model with Klingenberg CIs (Klingenberg, 2014) for the risk difference and the exact test of the null hypothesis that the common odds ratio is 1. Known predictors of treatment success were screened for their effects in addition to DYD and their interaction with DYD, while DYD was considered as log-concentration or dichotomized at the lower quartile. Repeated cycles were analyzed assuming some correlation between them for a given individual, namely by generalized estimating equations (GEEs) for prediction and generalized mixed models for an estimate of the variance component. A P-value of <0.05 was assumed to indicate statistical significance. All statistical analyses were performed by the software SPSS 27.0 (IBM, Chicago, IL, USA), GraphPad Prism 9.0 (GraphPad Software, Inc., San Diego, CA, USA) or R 4.0.3 (The R Foundation for Statistical Computing, Vienna, Austria).
Results
Patient flow and demographics
In total, N = 217 patients were included into this study. In n = 14 patients, a serum progesterone level >1.0 ng/ml on day of ET was detected, indicative of escape ovulation, and these patients were post hoc excluded from the analysis so as not to confound the association of DYD and DHD plasma levels obtained with 10 mg (tid) oral DYD and outcome by the presence of a corpus luteum. In n = 5 patients, no LC-MS/MS results were available for technical reasons, leaving n = 198 observations (n = 41 cleavage stage ETs and n = 157 blastocyst transfers) for the analyses. Mean (±SD) age of included patients was 33.2 ± 4.0 years, mean duration of infertility 42.6 ± 15.6 months and mean anti-Müllerian hormone concentrations 4.0 ± 2.7 ng/ml. An extensive description of patient demographics is shown in Supplementary Table SI. N = 56 patients had a second sample and n = 13 patients a third sample for analysis available, etc. The patient flow is shown in detail in Supplementary Fig. S2. In the first FET cycle, the overall ongoing pregnancy rate and live birth rate was 23% (45/198, 95% CI: 17% to 29%).
DYD and DHD plasma concentrations
Median (quartiles) of plasma levels of DYD and DHD were 1.36 ng/ml (0.738 to 2.17 ng/ml) and 34.0 ng/ml (19.85 to 51.65 ng/ml) on Day 2 or 3 and 1.04 ng/ml (0.707 to 1.62 ng/ml) and 30.0 ng/ml (20.8 to 43.3 ng/ml) on Day 5, respectively, suggesting that steady-state is reached already on Day 3 of oral DYD intake. Empirical distributions of DYD, DHD, progesterone and estradiol with distribution functions of lognormal distributions overlayed are Supplementary Fig. S1. Levels of DYD and its active metabolite DHD were strongly correlated (correlation coefficients of logarithms 0.942 on D2/3 and 0.931 for D5, respectively) irrespective of duration of DYD intake, which makes the allocation of DYD and DHD quarters similar (Fig. 1). Serum estradiol levels and DYD/DHD plasma showed no significant correlation (correlation coefficients of logarithms r = 0.085) for DYD and a similar, small, statistically significant correlation (r = 0.13) for DHD.
Double logarithmic scatterplot of dydrogesterone and dihydrodydrogesterone levels shows a strong correlation. Line, principal component regression.
Body weight and BMI
Linear regression from log DYD on log BMI showed slope −0.31 (CI: −0.86 to 0.21), i.e. practically no association (R2 = 0.007), while log DHD had slope −0.53 (CI: −1.02 to −0.07), i.e. a statistically significant effect of less DHD for a high BMI in the presence of a very weak association (R2 = 0.026). A scatterplot is shown in Fig. 2. The log values of body weight, too, had such weak associations (R2 = 0.020 for DYD and R2 = 0.043 for DHD) with log-concentrations, the slopes being −0.53 (CI: −1.07 to −0.01) for DYD and −0.68 (CI: −1.15 to −0.22) for DHD.
Logarithms of BMI (A) and bodyweight (B) are weakly associated with logarithms of plasma levels of dydrogesterone (squares) and dihydrodydrogesterone (circles).
Pregnancy to live birth achievement by DYD and DHD quarter
The 25th percentiles of DYD and DHD levels were 0.738 ng/ml and 19.85 ng/ml for cleavage stage transfers and 0.707 ng/ml and 20.80 ng/ml for blastocyst transfers, respectively. A comparison of the proportion of patients achieving a positive pregnancy test reveals a tendency toward a decreased chance to achieve a positive test in the lowest quarter of DYD levels (16/49 (33%) versus >25th percentile DYD 74/149 (50%), adjusted difference −17%; 95% CI: −32 to −1%, P = 0.047), which reached statistical significance for the cleavage stage subgroup (P = 0.028) (Fig. 3). In line with this finding, a reduction in the likelihood for an ongoing pregnancy was observed overall for patients in the lowest quarter of DYD levels as it was found for the chance to achieve a live birth (3/49 (6%) versus >25th percentile DYD 42/149 (28%), adjusted difference −22.0%, 95% CI: −31 to −11, P < 0.0008) (Tables I and II). Figure 4 shows a smooth spline function of the association of DYD and DHD levels and ongoing pregnancy including the dichotomization at the lower quartile. The same analysis for DHD quarters gave very similar results (Supplementary Tables SII and SIII, Fig. S3). The estimation algorithm of a three-parameter logistic regression did not converge to specific estimates. Analysis of pictorial blood assessment chart (PBAC) scores shows neither an association of DYD/DHD levels and bleeding patterns nor of PBAC results and cycle outcomes (data not shown).
Forest plot of risk differences with 95% CIs between lower quarter of dydrogesterone (DYD) levels and higher levels for different outcomes and embryo transfer at cleavage stage (D2/3) or as blastocyst (D5).
Probability of ongoing pregnancy as a function of logarithm of dydrogesterone (A) and dihydrodydrogesterone (B) levels at embryo transfer, observations, smooth spline with confidence band and the prespecified dichotomization at the lower quartile.
Outcomes of different dydrogesterone quarters with absolute numbers and relative proportion.
| Endpoint | Dydrogesterone <25th percentile Count (%) | Dydrogesterone >25th percentile Count (%) | Difference (%) (95% CI) |
|---|---|---|---|
| Positive pregnancy test | 16/49 (33) | 74/149 (49.7) | −17 (−31.3 to −0.9) |
| D2/3 | 1/9 (11) | 16/32 (50) | −38.9 (−59.9 to −2.2) |
| D5 | 15/40 (38) | 58/117 (49.6) | −12.1 (−28.5 to 5.9) |
| Preclinical pregnancy loss given a positive pregnancy test | 5/16 (31) | 16/74 (21.6) | 9.6 (−10.9 to 35.7) |
| D2/3 | 0/1 (0) | 2/16 (12.5) | −12.5 (−36.9 to 70) |
| D5 | 5/15 (33) | 14/58 (24.1) | 9.2 (−13.4 to 36.5) |
| Clinical pregnancy | 11/49 (22) | 58/149 (38.9) | −16.5 (−29.2 to −1.1) |
| D2/3 | 1/9 (11) | 14/32 (43.8) | −32.6 (−54 to 3.9) |
| D5 | 10/40 (25) | 44/117 (37.6) | −12.6 (−27.1 to 4.8) |
| Miscarriage 1st trimester given a positive pregnancy test | 8/16 (50) | 16/74 (21.6) | 28.4 (3.8 to 52.5) |
| D2/3 | 1/1 (100) | 4/16 (25) | 75 (−10.5 to 90.1) |
| D5 | 7/15 (47) | 12/58 (20.7) | 26 (0.8 to 51.6) |
| Ongoing Pregnancy | 3/49 (6) | 42/149 (28.2) | −22.1 (−31.2 to −10.1) |
| D2/3 | 0/9 (0) | 10/32 (31.2) | −31.2 (−48.8 to 1.3) |
| D5 | 3/40 (8) | 32/117 (27.4) | −19.9 (−30.4 to −5.8) |
| Live birth | 3/49 (6) | 42/149 (28.2) | −22.1 (−31.2 to −10.1) |
| D2/3 | 0/9 (0) | 10/32 (31.2) | −31.2 (−48.8 to 1.3) |
| D5 | 3/40 (8) | 32/117 (27.4) | −19.9 (−30.4 to −5.8) |
| Endpoint | Dydrogesterone <25th percentile Count (%) | Dydrogesterone >25th percentile Count (%) | Difference (%) (95% CI) |
|---|---|---|---|
| Positive pregnancy test | 16/49 (33) | 74/149 (49.7) | −17 (−31.3 to −0.9) |
| D2/3 | 1/9 (11) | 16/32 (50) | −38.9 (−59.9 to −2.2) |
| D5 | 15/40 (38) | 58/117 (49.6) | −12.1 (−28.5 to 5.9) |
| Preclinical pregnancy loss given a positive pregnancy test | 5/16 (31) | 16/74 (21.6) | 9.6 (−10.9 to 35.7) |
| D2/3 | 0/1 (0) | 2/16 (12.5) | −12.5 (−36.9 to 70) |
| D5 | 5/15 (33) | 14/58 (24.1) | 9.2 (−13.4 to 36.5) |
| Clinical pregnancy | 11/49 (22) | 58/149 (38.9) | −16.5 (−29.2 to −1.1) |
| D2/3 | 1/9 (11) | 14/32 (43.8) | −32.6 (−54 to 3.9) |
| D5 | 10/40 (25) | 44/117 (37.6) | −12.6 (−27.1 to 4.8) |
| Miscarriage 1st trimester given a positive pregnancy test | 8/16 (50) | 16/74 (21.6) | 28.4 (3.8 to 52.5) |
| D2/3 | 1/1 (100) | 4/16 (25) | 75 (−10.5 to 90.1) |
| D5 | 7/15 (47) | 12/58 (20.7) | 26 (0.8 to 51.6) |
| Ongoing Pregnancy | 3/49 (6) | 42/149 (28.2) | −22.1 (−31.2 to −10.1) |
| D2/3 | 0/9 (0) | 10/32 (31.2) | −31.2 (−48.8 to 1.3) |
| D5 | 3/40 (8) | 32/117 (27.4) | −19.9 (−30.4 to −5.8) |
| Live birth | 3/49 (6) | 42/149 (28.2) | −22.1 (−31.2 to −10.1) |
| D2/3 | 0/9 (0) | 10/32 (31.2) | −31.2 (−48.8 to 1.3) |
| D5 | 3/40 (8) | 32/117 (27.4) | −19.9 (−30.4 to −5.8) |
Bold face confidence limits indicate statistical significance.
D, day.
Outcomes of different dydrogesterone quarters with absolute numbers and relative proportion.
| Endpoint | Dydrogesterone <25th percentile Count (%) | Dydrogesterone >25th percentile Count (%) | Difference (%) (95% CI) |
|---|---|---|---|
| Positive pregnancy test | 16/49 (33) | 74/149 (49.7) | −17 (−31.3 to −0.9) |
| D2/3 | 1/9 (11) | 16/32 (50) | −38.9 (−59.9 to −2.2) |
| D5 | 15/40 (38) | 58/117 (49.6) | −12.1 (−28.5 to 5.9) |
| Preclinical pregnancy loss given a positive pregnancy test | 5/16 (31) | 16/74 (21.6) | 9.6 (−10.9 to 35.7) |
| D2/3 | 0/1 (0) | 2/16 (12.5) | −12.5 (−36.9 to 70) |
| D5 | 5/15 (33) | 14/58 (24.1) | 9.2 (−13.4 to 36.5) |
| Clinical pregnancy | 11/49 (22) | 58/149 (38.9) | −16.5 (−29.2 to −1.1) |
| D2/3 | 1/9 (11) | 14/32 (43.8) | −32.6 (−54 to 3.9) |
| D5 | 10/40 (25) | 44/117 (37.6) | −12.6 (−27.1 to 4.8) |
| Miscarriage 1st trimester given a positive pregnancy test | 8/16 (50) | 16/74 (21.6) | 28.4 (3.8 to 52.5) |
| D2/3 | 1/1 (100) | 4/16 (25) | 75 (−10.5 to 90.1) |
| D5 | 7/15 (47) | 12/58 (20.7) | 26 (0.8 to 51.6) |
| Ongoing Pregnancy | 3/49 (6) | 42/149 (28.2) | −22.1 (−31.2 to −10.1) |
| D2/3 | 0/9 (0) | 10/32 (31.2) | −31.2 (−48.8 to 1.3) |
| D5 | 3/40 (8) | 32/117 (27.4) | −19.9 (−30.4 to −5.8) |
| Live birth | 3/49 (6) | 42/149 (28.2) | −22.1 (−31.2 to −10.1) |
| D2/3 | 0/9 (0) | 10/32 (31.2) | −31.2 (−48.8 to 1.3) |
| D5 | 3/40 (8) | 32/117 (27.4) | −19.9 (−30.4 to −5.8) |
| Endpoint | Dydrogesterone <25th percentile Count (%) | Dydrogesterone >25th percentile Count (%) | Difference (%) (95% CI) |
|---|---|---|---|
| Positive pregnancy test | 16/49 (33) | 74/149 (49.7) | −17 (−31.3 to −0.9) |
| D2/3 | 1/9 (11) | 16/32 (50) | −38.9 (−59.9 to −2.2) |
| D5 | 15/40 (38) | 58/117 (49.6) | −12.1 (−28.5 to 5.9) |
| Preclinical pregnancy loss given a positive pregnancy test | 5/16 (31) | 16/74 (21.6) | 9.6 (−10.9 to 35.7) |
| D2/3 | 0/1 (0) | 2/16 (12.5) | −12.5 (−36.9 to 70) |
| D5 | 5/15 (33) | 14/58 (24.1) | 9.2 (−13.4 to 36.5) |
| Clinical pregnancy | 11/49 (22) | 58/149 (38.9) | −16.5 (−29.2 to −1.1) |
| D2/3 | 1/9 (11) | 14/32 (43.8) | −32.6 (−54 to 3.9) |
| D5 | 10/40 (25) | 44/117 (37.6) | −12.6 (−27.1 to 4.8) |
| Miscarriage 1st trimester given a positive pregnancy test | 8/16 (50) | 16/74 (21.6) | 28.4 (3.8 to 52.5) |
| D2/3 | 1/1 (100) | 4/16 (25) | 75 (−10.5 to 90.1) |
| D5 | 7/15 (47) | 12/58 (20.7) | 26 (0.8 to 51.6) |
| Ongoing Pregnancy | 3/49 (6) | 42/149 (28.2) | −22.1 (−31.2 to −10.1) |
| D2/3 | 0/9 (0) | 10/32 (31.2) | −31.2 (−48.8 to 1.3) |
| D5 | 3/40 (8) | 32/117 (27.4) | −19.9 (−30.4 to −5.8) |
| Live birth | 3/49 (6) | 42/149 (28.2) | −22.1 (−31.2 to −10.1) |
| D2/3 | 0/9 (0) | 10/32 (31.2) | −31.2 (−48.8 to 1.3) |
| D5 | 3/40 (8) | 32/117 (27.4) | −19.9 (−30.4 to −5.8) |
Bold face confidence limits indicate statistical significance.
D, day.
Risk differences (RDs) between lowest dydrogesterone quarter and the rest stratified by day of embryo transfer with Klingenberg CI and exact P-value.
| Endpoint | Adj. RD (%) | 95% CI | P-value |
|---|---|---|---|
| Positive pregnancy test | −17 | −32 to −1 | 0.0471 |
| Preclinical pregnancy loss given a positive pregnancy test | +8 | −16 to 34 | 0.531 |
| Clinical pregnancy | −16 | −30 to −1 | 0.0396 |
| Miscarriage 1st trimester given a positive pregnancy test | +30 | 2.3 to 55 | 0.0259 |
| Ongoing pregnancy | −22 | −31 to −11 | 0.000808 |
| Live birth | −22 | −31 to −11 | 0.000808 |
| Endpoint | Adj. RD (%) | 95% CI | P-value |
|---|---|---|---|
| Positive pregnancy test | −17 | −32 to −1 | 0.0471 |
| Preclinical pregnancy loss given a positive pregnancy test | +8 | −16 to 34 | 0.531 |
| Clinical pregnancy | −16 | −30 to −1 | 0.0396 |
| Miscarriage 1st trimester given a positive pregnancy test | +30 | 2.3 to 55 | 0.0259 |
| Ongoing pregnancy | −22 | −31 to −11 | 0.000808 |
| Live birth | −22 | −31 to −11 | 0.000808 |
Risk differences (RDs) between lowest dydrogesterone quarter and the rest stratified by day of embryo transfer with Klingenberg CI and exact P-value.
| Endpoint | Adj. RD (%) | 95% CI | P-value |
|---|---|---|---|
| Positive pregnancy test | −17 | −32 to −1 | 0.0471 |
| Preclinical pregnancy loss given a positive pregnancy test | +8 | −16 to 34 | 0.531 |
| Clinical pregnancy | −16 | −30 to −1 | 0.0396 |
| Miscarriage 1st trimester given a positive pregnancy test | +30 | 2.3 to 55 | 0.0259 |
| Ongoing pregnancy | −22 | −31 to −11 | 0.000808 |
| Live birth | −22 | −31 to −11 | 0.000808 |
| Endpoint | Adj. RD (%) | 95% CI | P-value |
|---|---|---|---|
| Positive pregnancy test | −17 | −32 to −1 | 0.0471 |
| Preclinical pregnancy loss given a positive pregnancy test | +8 | −16 to 34 | 0.531 |
| Clinical pregnancy | −16 | −30 to −1 | 0.0396 |
| Miscarriage 1st trimester given a positive pregnancy test | +30 | 2.3 to 55 | 0.0259 |
| Ongoing pregnancy | −22 | −31 to −11 | 0.000808 |
| Live birth | −22 | −31 to −11 | 0.000808 |
Preclinical pregnancy loss and miscarriage
The proportion of patients having a preclinical pregnancy loss between patient groups with different DYD/DHD levels was of a comparable magnitude (≤25th percentile 31%, versus >25th percentile 22%, adjusted risk difference +8%, 95% CI: −16% to 34% for DYD). As these frequencies are reported for the subgroup with a positive pregnancy test, the investigation was not powered for a narrow CI. For the risk to suffer a miscarriage, however, a significant increase in miscarriages for patients in the lowest quarter of DYD/DHD levels was observed overall (50%, versus 22%, adjusted risk difference +30%, 95% CI: –2 to 55% for DYD) (Table II and Supplementary Table SIII).
Regression analysis
Multiple logistic regressions revealed plasma DYD/DHD levels below versus above the 25th percentile as an independent prognostic factor for decreasing the live birth rate in the analyzed study population, as adjusting for each confounding variable did not change the estimated odds ratio (OR) significantly. The OR to achieve a live birth or an ongoing pregnancy for patients in the lowest quarter was 0.16 (95% CI: 0.5 to 0.04) for DYD and 0.23 (95% CI: 0.62 to 0.07) for DHD levels on day of ET.
Only when estradiol levels were considered, a significant interaction was estimated (Fig. 5), which was the only event that the log-concentration of DYD was estimated to have a significant effect upon live birth rate. The P-value of 0.0165 would, however, not be considered significant when adjusting for multiple testing. The observations that generate this new hypothesis are the occurrence of three pregnancies despite DYD <0.7 ng/ml (in the lower quarter) in women with estradiol >260 pg/ml (in the upper quarter).
Hormone levels at embryo transfer for 149 first cycles without ongoing pregnancy (circles) 49 ongoing pregnancies (filled circles) with color coding the estimated probability of an ongoing pregnancy in a logistic regression on logarithms with interaction. There are no ongoing pregnancies in the lower left part with dydrogesterone <0.711 ng/ml (lower quarter) at least for estradiol <263 pg/ml (upper quarter).
Repeated measurements
Analysis of repeated cycles with GEE for plasma DYD/DHD levels below versus above the 25th percentile led to very similar results, e.g. a relevant interaction of DYD and estradiol (Supplementary Fig. S4). When the effect of cycle being the first was investigated, the interaction with DYD quarter eliminated half the effect of high DYD in subsequent cycles. Predicted (observed) rates of ongoing pregnancy were 6% (3/49) in the first cycle and 6% (1/16) in subsequent cycles in the lower DYD quarter and 28% (42/149) in the first and 16% (10/60) in further cycles with higher DYD. These results are very preliminary, as there was just one ongoing pregnancy in the lower DYD quarter in a third cycle. The variance component for the individual intercepts was estimated to be 1.07 in a similar model that considered cycles 1–2 versus 3+ in order to be estimable.
When repeated measurements in subsequent FET cycles of logarithms of DYD levels were explained by individual patient as random effects and cycle as fixed effect, variance components were 0.2922 between patients and 0.552 within patients, so that the intraclass correlation coefficient is 0.22. The respective values for DHD are 0.2572, 0.5012 and 0.209. The coefficient of variation within patients is 60% of the individual mean DYD level, while the CV between patients is just 30% of the population mean DYD level. Supplementary Fig. S5 depicts individual DYD profiles across subsequent cycles.
Obstetrical and perinatal outcomes
For the course of the pregnancy, no difference in pregnancy associated complications such as pregnancy-induced hypertension, pre-eclampsia/HELLP-syndrome or gestational diabetes were found between lower and higher DYD/DHD patients (data not shown). There were no perinatal death or stillbirth in the study population, mean birth weight was similar between patients in the group ≤25th versus >25th of DYD/DHD levels on day of ET (3942 ± 520 g, 95% CI: 2650 to 5233 versus 3599 ± 516 g, 95% CI: 3424 to 3773).
Discussion
The use of FET treatment cycles has seen a strong increase over the past decade (Assisted Reproductive Technology Fertility Clinic Success Rates Report, 2018). The programmed artificial FET cycle is widely used for that purpose, under the assumption of at least equivalent efficacy in terms of pregnancy achievement as compared to a natural cycle or modified natural cycle (Ghobara et al., 2017). Furthermore, an artificial FET regimen comes with the advantage of an easy alignment of the time point of thawing and transferring embryos with organizational necessities of the IVF laboratory, the treating doctors and the patient. Herein, we explore the use of DYD as the progestogenic drug for inducing endometrial receptivity and supporting early pregnancy up to the luteo-placental shift in an artificial cycle (Neumann et al., 2020a,b).
DYD was first introduced to the market in 1961 and has since been used, on a global scale, in a number of indications in women’s health, such as menstrual disorders and threatened or recurrent miscarriage (Griesinger et al., 2019). Recently, the LOTUS trial program (Tournaye et al., 2017; Griesinger et al., 2018) has renewed the interest in using DYD for LPS in IVF. The licensed dose of DYD for LPS in a fresh IVF cycle is 10 mg (tid) showing good efficacy when compared against standard-of-care MVP treatments (Griesinger et al., 2020; TEGGO (The Eshre Guideline Group On Ovarian Stimulation), 2020). Of note, the 10 mg tid DYD dose now licensed for LPS in fresh IVF cycles in several countries has been developed empirically. This is indeed the case for all LPS drug regimen, which have been tested only for surrogate outcomes such as serum and tissue levels and/or endometrial transformation within small dose-finding studies (de Ziegler et al., 2013; Paulson et al., 2014; Duijkers et al., 2018). Although key aspects of DYD pharmacokinetics are known (Kuhl, 2005; Griesinger et al., 2019), plasma levels at steady state of 10 mg (tid) have not been published. Neither has the inter- and intraindividual variation in DYD plasma levels been described, nor the potential interindividual variation in the metabolism of DYD to DHD, the main active metabolite (Olbrich et al., 2016) been assessed in a real-world setting. For assessing the association of DYD and DHD levels with outcome, the artificial FET cycle is optimal. DYD does not cross-react with bioidentical progesterone in ELISA or LC/MS assay (Neumann et al., 2020a,b). Being able to distinguish endogenous progesterone from DYD exposure and thereby excluding escape ovulations from the analysis is one strength of the present study, as it allows an unbiased estimate of DYD/DHD plasma levels necessary for pregnancy achievement and maintenance.
The present study confirms for the 10 mg tid dosage that steady-state in plasma is achieved rapidly after oral administration. With the 10 mg DYD (tid) dosage, we observed median plasma levels of 1.36 ng/ml DYD and 34.0 ng/ml DHD on the third day of intake and similar levels thereafter, closely resembling the DYD: DHD ratio of 1:25 that has previously been reported (Rižner et al., 2011). Given the strong correlation of DYD and DHD levels, our data further suggest that in future clinical studies on the association of DYD/DHD exposure with outcome, only one of the two compounds needs to be assessed by the elaborate and costly LC/MS method. Given the fact that 20α-DHD is considered the active metabolite of the drug, it could be advisable to concentrate future efforts on the study of DHD levels. It is noteworthy, however, that little is known about the genetic variation between populations and individuals in the activity of the aldo-keto reductase 1C (AKR1C) enzyme family members and cytochrome P450 (CYP) isoenzymes, the main drivers in the metabolic pathway from DYD to 20α-DHD (Olbrich et al., 2016; Alshogran, 2017). Accordingly, the conversion of DYD to 20α-DHD may differ in rare individuals or in distinct populations other than Caucasians.
This study reports considerable interindividual variation of DYD/DHD blood concentrations, with 25-fold differences in plasma levels between individuals at the maximum. Body weight and body mass index as proxies for volume of distribution cannot explain this variation. This is, however, in line with the observation that the dosing of other orally administered sex steroids, such as estrogens, has little or no association with serum blood levels, a lack of or only weak dose proportionality and no association with body mass index (Nolan et al., 2020; Kim et al., 2021). Given the fact that oral bioavailability for sex steroids is generally low (Kuhl, 2005), small variations in absorption can have a strong impact on blood concentrations. Another potential confounder of DYD/DHD blood concentrations is concomitant food intake, which may affect not only immediate intestinal absorption itself but also the enterohepatic circulation, a phenomenon where a compound is excreted via bile into the small intestine and then reabsorbed and excreted into bile again (Vinarov et al., 2021). For oral testosterone, for example, a more than 10-fold difference in mean serum levels was observed in patients in a fasting state versus patients having the compound administered together with a standardized meal (Bagchus et al., 2003). For oral micronized progesterone, a 2-fold enhanced absorption was observed in the presence of food intake versus fasting (Simon et al., 1993). It is noteworthy, however, that estradiol levels and DYD/DHD levels exhibited nearly no correlation in the present study, which does not support the hypothesis that differences in food intake could explain the DYD/DHD plasma level variation observed (assuming same effect direction of food intake for estradiol valerate and DYD). Another effect that comes potentially into play is that oral administration of sex steroids such as estrogen and progesterone will elicit pulses in the serum concentration profile even at steady-state (Setnikar et al., 1996; Wang et al., 2019). Such short pulses of approximately 3–4 h after dosing have recently also been reported for oral DYD 10 mg (tid) administration in a study on the luteal phase characteristics of oocyte donors (Mackens et al., 2021). Since ETs in our unit are routinely not done before 11.00 a.m., we speculate that the pulses after the morning intake at breakfast time will not have contributed significantly to the variability of DYD/DHD blood concentrations observed in the present study. However, in future studies, food intake and timing of dosing/sampling should be controlled or at least recorded. Another noteworthy observation from the present study is the intraindividual variability between subsequent FET cycles in DYD/DHD concentrations on day of FET, suggesting that differences in plasma levels are less determined by inherent patient characteristics, but rather the consequence of the very circumstances and conditions of the individual FET cycle.
The present data also suggest that the blood level threshold of DYD and DHD levels to achieve good efficacy is ∼0.7 ng/ml and 20 ng/ml, respectively on day of FET. However, this observation should be replicated in further and larger clinic cohorts to verify the external validity of our finding. The present trial findings also generate the hypothesis that in an artificial FET cycle, likewise to a natural cycle (Romanski et al., 2021), progestogenic exposure must not be seen in isolation from estrogenic exposure, since we observed a remarkable interaction of low estradiol levels with low DYD/DHD levels and outcome.
This present trial was prompted by the observation published in 2017 of low progesterone levels in a quarter of subjects treated with high doses of MVP in an artificial cycle and the associated reduction in the ongoing pregnancy rate of a magnitude of ∼20% (Labarta et al., 2017). For this reason, we abandoned the routine use of MVP in artificial programmed FET cycles in our center and turned our interest to the convenient, orally active and systemically acting compound DYD. The present analysis, however, indicates that the use of DYD 10 mg (tid) as the sole progestogenic drug could be suboptimal for a proportion of the patient population, and the need for further research is evident. Given the increasing safety concerns about artificial FET cycles in which iatrogenically ovulation is suppressed and a corpus luteum is absent in early pregnancy (von Versen-Höynck et al., 2019; Conrad et al., 2021; Pereira et al., 2021) it could be worthwhile to concentrate further research efforts on studying the use of progestogenic drugs in the context of an ovulatory cycle. This is underlined by the rather low overall ongoing pregnancy chance of 23% in this study which is likely to result in part from progesterone underexposure in a quarter of the population. Still, for ovarectomized women, women with age-related ovarian depletion or for oocyte donation recipients needing cycle synchronization, the optimization of artificial programmed FET cycle regimen is mandatory (Labarta and Rodríguez, 2020; Labarta et al., 2021). As corroborated herein, escape ovulations do occur under 6 mg estradiol valerate treatment (Dal Prato et al., 2002), and, assuming they remain often undetected, it may be speculated that the association of anovulatory cycle and adverse events in pregnancy (Ginström Ernstad et al., 2019) is underestimated in many of the existing observational and registry studies.
In conclusion, plasma levels of DYD and its metabolite DHD vary considerably between and within individuals in programmed FET cycles using daily 10 mg oral DYD (tid) in the investigated cohort of patients. These variations are independent from body weight or BMI. Lower plasma levels of DYD and DHD are associated with a lower likelihood for an ongoing pregnancy. Daily 10 mg DYD (tid) in an artificial FET cycle is therefore potentially a suboptimal dose for a proportion of the population. Future studies can build on the present data, in order to optimize outcome (e.g. explore higher DYD dosages or a combination of DYD with another progestogenic drug).
Supplementary data
Supplementary data are available at Human Reproduction online.
Data availability
The data underlying this article will be shared on reasonable request to the corresponding author.
Acknowledgements
Our thanks go to Mrs Andrea Knaak, study nurse at the Department of Gynecological Endocrinology and Reproductive Medicine, University Hospital of Schleswig-Holstein, Campus Luebeck, Luebeck, Germany, for document management, data entry, sample management and sample shipment, as well as pregnancy and neonate follow-up. We also thank Peter van Amsterdam, Abbott Healthcare Products B.V., the Netherlands, for his input on DYD LC-MS/MS measurements and thoughtful insights on pharmacokinetics.
Authors’ roles
K.N. was responsible for study initiation, study conduct and supervision, data analysis and interpretation, manuscript drafting and finalizing. A.M. was responsible for data acquisition from CRFs and patients’ files, data analyses and interpretation. R.V. was responsible for data analyses, interpretation, manuscript drafting and finalization. G.G. was responsible for study design, study initiation and supervision, data analysis and interpretation, manuscript drafting and finalizing. M.D., T.K.E. and A.S.-M. were investigators and critically reviewed the analyses and the manuscript through several revision rounds.
Funding
The trial was financed by university funds, except for the cost for plasma and serum sample handling, storage and shipment, as well as the liquid chromatography–mass spectrometry (LC-MS) analysis of DYD, DHD and progesterone, which was financially supported by Abbott Products Operations AG (Allschwil, Switzerland). Abbott Products Operations AG had no influence on study protocol, study conduct, data analysis or data interpretation.
Conflict of interest
K.N. has received honoraria and/or non-financial support (e.g. travel cost compensation) from Ferring, Gedeon-Richter, Merck and MSD. A.M. has no competing interests. R.V. has no competing interests. M.D. has received honoraria and/or non-financial support from Ferring and Merck. A.S.-M. has no competing interests. T.K.E. has received honoraria and/or non-financial support from Roche, Novartis, Pfizer, Aristo Pharma, Merck. G.G. has received honoraria and/or non-financial support (e.g. travel cost compensation) from Abbott, Ferring, Gedeon Richter, Guerbet, Merck, Organon, MSD, ObsEva, PregLem, ReprodWissen GmbH, Vifor and Cooper.
References
Assisted Reproductive Technology Fertility Clinic Success Rates Report.
TEGGO (The Eshre Guideline Group On Ovarian Stimulation),
