A model for predicting low progesterone levels on the day of embryo transfer in hormonally prepared vitrified‐warmed embryo transfer cycles: A secondary analysis of a prospective cohort study

To develop a model that predicts low progesterone (P) levels on the day of embryo transfer (ET) based on patient and cycle characteristics, including serum estradiol (E2) concentration after vaginal administration of micronized E2 for endometrial preparation.


INTRODUCTION
Hormone replacement therapy (HRT) cycles are among the most common in vitro fertilization protocols for priming the endometrium prior to frozen-thawed embryo transfer (FET).Despite the worldwide use of this protocol, the optimal route of administration and dosing of progesterone (P) remains unknown.
The first published study on this topic has demonstrated an association between P levels and reproductive outcomes in HRT cycles. 1 Subsequently, several studies have reported similar results, [1][2][3][4][5][6][7][8][9][10] reaching the same conclusion that a threshold serum P level must be reached to optimize the reproductive outcome of an HRT cycle.However, there is a lack of evidence to explain the significant inter-patient differences in serum P values after vaginal P administration.Moreover, whether these variations are caused by intrinsic patient characteristics, cycle aspects, or other factors remains unclear.The differences in the clearance of drugs and, even more importantly, vaginal absorption rates are prone to affect serum P levels. 7iven the growing consensus that "one size does not fit all" in HRT, this study sought to identify patient and cycle characteristics that predict low serum P levels around the time of embryo transfer (ET) after vaginal P administration.Specifically, we hypothesized that the early estradiol (E 2 ) level after vaginal administration of micronized E 2 would provide information on the absorption rate of the vaginal epithelium and thus act as a predictor of the subsequent serum P level when P is also administered vaginally.

MATERIALS AND METHODS
This study was conducted at the Akdeniz University Hospital Assisted Reproduction Unit between December 2020 and April 2021.This is a secondary analysis of a prospective observational trial, that explores the effect of endometrial compaction (decreased endometrial thickness) on pregnancy rates in which patients were supplemented with micronized E 2 vaginally. 11Ethical approval was obtained from the local research ethics committee (Approval No. 2021-594).All programmed FET cycles with highquality blastocysts were included.No age limit was applied when recruiting patients.All natural, modified natural, and HRT cycles with preimplantation genetic testing were excluded from the analysis.Each patient was included in the study only once.All patients gave informed consent, and patient anonymity was preserved.
All examined cycles included exogenous hormonal preparation of the endometrium.No down-regulation with gonadotropin-releasing hormone agonists was used.All patients received the same FET protocol, starting hormone therapy on day 3 or 4 of a natural cycle, using vaginal administration of 2 mg micronized E 2 twice daily (Estrofem, Novo Nordisk, Turkey).On day 5 of vaginal E 2 administration, a transvaginal ultrasound scan was performed with a GE P6 (GE Healthcare, Seoul, Korea), using a wideband 5-to 9-MHz endocavitary transducer to measure the endometrial thickness.After the scan, blood samples were collected to measure the serum E 2 levels retrospectively.On day 5 (after blood sampling), E 2 administration was changed from vaginal to oral in all patients (Estrofem, Novo Nordisk, Turkey, 6 mg/day).After 10-12 days of E 2 administration once the endometrium was ≥7 mm, E 2 supplementation continued, and a regimen of 100 mg of vaginal micronized P tablets (Lutinus, Ferring GmbH, Germany) three times daily commenced.The patients were admitted to the clinic between 8:00 AM and 9:00 AM on day 6 of P treatment, where serum P levels were measured, and a transvaginal ultrasound was performed to determine the final endometrial thickness.If the serum P level on the day of ET was lower than 10 ng/mL, an additional 50 mg of IM-administered P (Progestan, Kocak Farma, Istanbul) was given every other day until the pregnancy test.In pregnant patients, daily E 2 and vaginal P were continued for luteal support until 10 weeks of gestation.All FET cycles were performed by the same physician and the laboratory investigations were performed at a single laboratory during the entire study period.
Two experienced embryologists evaluated all blastocysts before ET, according to the Alpha Istanbul Consensus guidelines, 12 grading them based on the level of expansion and the quality of the inner cell mass and trophectoderm.Day 5 embryos were assigned grades 2-4 based on blastocyst expansion and hatching status.A grade of A or B based on the inner cell mass and trophectoderm structure was considered high quality.
A positive pregnancy test was defined as a serum beta-human chorionic gonadotrophin level exceeding 10 IU/L 9 days after ET.Clinical pregnancy was defined as the visualization of a gestational sac via transvaginal ultrasound examination and/or documentation of trophoblastic tissue in a miscarriage specimen.Miscarriage was defined as a spontaneous loss of intrauterine pregnancy before 12 weeks of gestation, excluding those with biochemical pregnancy loss.Ongoing pregnancy was defined as pregnancy beyond 12 weeks of gestation.
The primary outcome measure is the serum P concentration measured on the morning of the ET day following vaginal P administration.
All data were analyzed using SPSS for Mac OS, version 26.0 (SPSS Inc., Chicago, IL, USA).Before analysis, we assessed data normality using visual methods such as histograms and probability plots, as well as the Kolmogorov-Smirnov test.The distributions of continuous variables are presented as the median and interquartile range (IQR), while categorical variables are given in percentages.Differences in continuous variables among groups were determined using the Mann-Whitney U test.Chi-square tests were used to compare categorical variables.Receiver operator characteristic (ROC) curve analysis was performed to evaluate the predictive values of patient weight and day 5 E 2 level for P level on the day of ET.Multivariate linear regression analysis was also performed, using variables that were significantly different between patients with low and adequate P levels on the day of ET.The Hosmer-Lemeshow test for the goodness of fit of the models was applied.The diagnostic indices (sensitivity and specificity), positive and negative predictive values (PPVs and NPVs), and likelihood ratios (LRs; positive and negative) were calculated for high weight, low day 5 E 2 level, and a combination of these two variables.In addition, the predictive performance of low day 5 E 2 on the day of ET was calculated among high-weight patients in a sequential model.p-value <0.05 was considered statistically significant.

RESULTS
A total of 193 FET cycles using vaginal micronized E 2 for endometrial preparation for 4 days were analyzed.
The median serum P level was 10.0 ng/mL (IQR 7.8-12.8ng/mL).When grouped by low (<7.8 ng/mL, at or below the 25th percentile) and adequate (≥7.8 ng/mL) serum P levels, patients with low levels were heavier ( p < 0.001), had higher body mass index (BMI; p < 0.001) and had a longer duration of infertility ( p = 0.045) compared with those with adequate P levels (Table 1).The cycle hormone levels are presented in Table 2. Baseline follicle-stimulating hormone (FSH) was higher ( p = 0.050) and day 5 serum E 2 levels were lower (p < 0.001) in the low-P group.
Multivariate linear regression analysis was used to assess the role of weight, BMI, day 5 serum E 2 levels, FSH levels, and infertility duration in predicting P levels on the day of ET after controlling for age.The results indicated that only weight ( p = 0.003) and day 5 E 2 level ( p < 0.001) were independently associated with P levels.
When only high weight or low day 5 E 2 were used to predict a low P level, 35 of these 49 (71.4%) were accurately diagnosed.However, 64 (64.6%) with high weight and 55 (61.1%) with low day 5 E 2 levels were inaccurately predicted to have an inadequate P level on the day of ET.When high weight and low day 5 E 2 were combined, 29 of the 49 patients (59.2%) with low P levels were accurately diagnosed (Table 3).Of the 99 patients with high weight, 35 (35.4%) had low P levels on the day of ET, and of the 90 patients with low day 5 E 2 levels, 35 (38.9%) had low P levels on the day of ET.The probability of a low P level on the day of ET increased from 25% to 35.4% for high-weight patients and to 38.9% for those with a low day 5 E 2 (Table 4).Of the 53 patients (27.5%) with both high weight and low day 5 E 2 levels, 29 (54.7%) had low P levels on the day of ET.The combination of these two factors increased the probability of a low serum P level on the day of ET from 25% to 54.7% (Table 4).

DISCUSSION
This study showed that low day 5 E 2 levels after vaginal administration of E 2 and high weight are independently associated with low P levels on the day of ET.Studies using either oral or transdermal  formulations of E2 for endometrial preparation have reported no association between the P level and serum E2 levels during the proliferative phase. 9,13However, similar to our results, Cédrin-Durnerin et al. (2019) observed that patients with P < 10 ng/mL had lower E2 levels than those with higher P levels following daily vaginal administration of 2 mg micronized E 2 . 7hysicochemical properties such as molecular weight, lipophilicity, ionization, surface charge, and chemical nature, can influence vaginal drug absorption.Accordingly, vaginal permeability is high for lipophilic steroids like estrogen and progesterone. 14However, any biological factor that affects drug dissolution and membrane transport (including cyclic changes in vaginal epithelium thickness, fluid volume and composition, pH, and sexual arousal) could potentially affect the vaginal absorption profile of the patient. 15,16Thus, vaginal administration of micronized E 2 may provide insight into the absorption rate of the vaginal epithelium and act as a predictor of the subsequent serum P level when P is also administered vaginally.
Several factors are related to the P level on the day of ET.[7]9 Our study found no significant age differences between patients with low and adequate serum P levels.Although some previous researchers did not find any significant association between P level and weight, 5,17 negative correlations between these variables have recently been reported. 1,6,9,13,18Our findings are in line with the latter studies, reporting that body weight is independently associated with the serum P level on the day of ET.
Predicting the serum P level is a novel perspective that may help individualize luteal phase support according to patient characteristics.Although the factors determining the serum P level on the day of ET have been previously investigated, none of these studies proposed a prediction model.The individual PPVs of high patient weight and low day 5 E 2 in predicting a low P level were found to be comparable.The highest PPV was observed in patients with both high weight and low day 5 E 2 .Interestingly, using either variable alone yielded similar NPVs as combining them.Therefore, a prediction based only on weight would be sufficient if the primary aim is to exclude low P levels on the day of ET.However, combining weight and day 5 E 2 increases the specificity and PPV and is, therefore, more appropriate for predicting which patients are at risk for a low P level on the day of ET.In addition, a low day 5 E 2 level improved the predictive performance of high weight with sequential use.This finding suggests that patient weight could be screened before initiating a FET cycle, and vaginal micronized E 2 testing may be used only in patients with a weight above 65 kg.Moreover, the highest sensitivity value (82.9%) was obtained for P < 7.8 ng/mL.Accordingly, such patients may be given higher P doses or a combination of P administration routes from the first day of supplementation.A P cut-off of 7.8 ng/mL (25th percentile) was chosen empirically.In fact, higher sensitivity values were observed if the critically low P level of 6.3 ng/mL (10th percentile) was used for each test or combination of tests.
Arguably the accuracy level would be inadequate if we solely rely on these parameters, and when progesterone measurement is available.However, the clinical value of progesterone measurement rests on the assumption that salvage treatments are truly effective.It should be kept in mind that data on the effects of a rescue bolus of P (either IM or SC) are not supported by robust evidence.0][21] Given that cycle cancellations due to extremely low levels of P still stand, 19 some patients may benefit from starting with higher doses or multiple administration routes of P administration earlier.
The use of the day 5 serum E 2 level after vaginal administration of micronized E 2 is advantageous because it yields rapid, objective, and quantitative results that are easy to interpret and are applicable in all clinics.However, the day 5 E 2 levels might vary in settings with different vaginal micronized E 2 preparations or dosages.Therefore the present cut-off level needs to be confirmed by others.Since the patients with low P levels (< 10.0 ng/mL) in this study were supplemented with IM-administered P from the day of ET onward, we did not aim to compare the reproductive outcomes according to the P levels.Accordingly, there was no significant difference in the ongoing pregnancy rates between patients with low and adequate P levels on the day of ET.][21] The limitation of the present study is the relatively moderate sample size.However, in the post-hoc analysis of 193 patients with five predictor variables, the test power (1 À β) at a 95% CI (1 À α) was calculated to be 99% using G*Power 3.1.9.7.Additionally, its strength is that a single clinician performed all of the FET cycles included in the analysis in a homogeneous study population under the same clinical settings and laboratory conditions.
In conclusion, the results of the present study suggest that a low day 5 E 2 level after vaginal administration of micronized E 2 and high weight are independently associated with low serum P levels on the day of ET and that the predictive performance of these variables is enhanced when they are used sequentially or in combination.We believe these findings may help clinicians individualize luteal phase support in HRT-FET cycles.Future research should be directed toward validating the present results.

T A B L E 1
Baseline characteristics of the patients according to low or adequate progesterone levels on the day of embryo transfer.P < 7.8 ng/mL P ≥ 7.8 ng/mL p (n = 49) (n = 144) Cycle characteristics of the patients according to low or adequate progesterone levels on the day of embryo transfer.
T A B L E 3 Predictive performance of single, combined, or sequential use of variables for predicting low progesterone levels on the day of embryo transfer.Frequency of low progesterone levels on the day of embryo transfer according to high weight and low day-5 estradiol level.