Effect of luteal phase support with progesterone on thyroid function: a sub-study of the FET optimizing trial

This work was supported by an unrestricted grant from Gedeon Richter (Forward Grant, FORWARD2018_5) and Rigshospitalet’s Research Foundation covering the study expenses. The funders of the study had no role in study design, data collection, data analysis, data interpretation, or writing of this report.

This study investigates the effect of LPS with vaginal progesterone supplementation on thyroid function in modified natural cycle (mNC) frozen–thawed embryo transfer (FET). The study is a sub-study of an RCT investigating the effect of progesterone LPS and the timing of blastocyst transfer in mNC FET on the live birth rate ( 11 ). The mNC-FET is characterized by the transfer of a frozen–thawed embryo in a natural menstrual cycle where the administration of a single dose of hCG (LH activity) is used to induce endometrial luteinization and endometrial–embryo synchronicity at the time of the embryo transfer. The luteal phase approximately corresponds to the period between the day of hCG trigger and the day of pregnancy testing. The main study was conducted between January 2019 and March 2024 at eight Danish public fertility clinics. The sub-study was conducted between January 2019 and October 2022 at Horsens Regional Hospital and Copenhagen University Hospital Rigshospitalet. The study protocol ( 12 ) and the reproductive outcomes of the RCT were previously published ( 11 ). Overall, the progesterone LPS and the timing of blastocyst transfer did not affect the live birth rate or other obstetric outcomes. We adhere to the 2025 CONSORT statement for the reporting of RCT ( 13 ). Participants were 18–41 years of age, had regular menstrual cycles (23–35 days), and underwent mNC-FET treatment with single blastocyst transfer derived after homologous IVF/ICSI. Detailed inclusion and exclusion criteria have already been described ( 12 ). All participants underwent postponed FET, meaning at least one full menstrual cycle passed between ovarian stimulation and oocyte retrieval and the subsequent FET cycle. In this sub-study, any known thyroid disease was an additional exclusion criterion, e.g. treatment with antithyroid drugs, supplementation with levothyroxine or triiodothyronine, or a baseline TSH level outside the reference range. All women participating in the main study were consecutively invited to participate in the sub-study. Participation required a separate informed consent and storage of blood samples in a biobank. Recruitment to the sub-study ended when at least 80 women had been randomized to LPS or no LPS. Participants were allocated to one of the four study groups (1:1:1:1): (A) blastocyst transfer day 6 following hCG trigger and progesterone LPS, (B) blastocyst transfer day 7 and progesterone LPS, (C) blastocyst transfer day 6 and no LPS, and (D) blastocyst transfer day 7 and no LPS. The randomization was performed on the day of hCG trigger or the day before and was conducted by research staff, using a computerized program running a minimization algorithm (Doxos CDS, Sweden). The randomization algorithm has previously been described ( 12 ). Women attended a screening visit on cycle days 2–5 where screening blood samples were drawn. The included women were followed with transvaginal ultrasonography from the mid-follicular phase until the dominant follicle reached ≥17 mm (day 0) when the ovulation trigger, 250 μg hCG (Ovitrelle®, Merck), was administered at 10 pm. Women allocated to progesterone LPS received vaginal suppositories with progesterone (100 mg*3/daily, Lutinus®, Ferring Pharmaceuticals, Denmark) starting on day 4 following hCG trigger. A single blastocyst transfer was performed on day 6 or 7 after ovulation trigger (according to randomization group). If the blastocyst did not survive thaw and no other blastocyst was available, treatment was canceled. At day 16, following the hCG trigger, a pregnancy test (s-hCG) was conducted. An s-hCG level ≥5 IU/L was considered a positive test. When s-hCG was between 5 and 49 IU/L, patients were followed as per local instruction with repeated s-hCG measurements until a sufficient increase in s-hCG or a fall in s-hCG level <5 IU/L (pregnancy loss) was confirmed. Patients with an s-hCG level <5 IU/L were considered non-pregnant. Blood samples were drawn at screening, day of hCG trigger (day 0), day of blastocyst transfer (6- or 7-days following ovulation trigger), and days 11, 14, 16, and 19 after hCG trigger. The primary outcome is the change in TSH during the luteal phase in women receiving or not receiving progesterone LPS (study group A + B vs study group C + D). Secondary outcomes include the change in fT4 and serum (s-) progesterone between groups. At study site 1 (Copenhagen University Hospital Rigshospitalet, Copenhagen, Denmark), TSH, fT4, and s-progesterone were measured by electrochemiluminescence immunoassays (Cobas 8000, e801 Roche Diagnostic, Germany). The detection level for TSH was 0.01 IU/L with a CV max of 4% at both 2 × 10 −3 IU/L and 6 × 10 −3 IU/L. For fT4, the detection level was 1.3 pmol/L with a CV max of 6% at both 15 and 40 pmol/L. For s-progesterone, the detection level was 0.6 nmol/L with a CV max of 10% at 3 nmol/L. Thyroid peroxidase antibody (TPOab) and thyroglobulin antibody (TGab) were analyzed using an immunofluorometric assay (B.R.A.H.M.S Kryptor Compact PLUS). For TPOab, the detection level was 50 ie/L and CV max at 140 and 260 ie/L was 13 and 8%, respectively. For TGab, the detection level was 33 ie/L and CV max at 100 and 340 ie/L was 7 and 5%, respectively. At study site 1, a TPOab value ≥60 ie/L and a TGab value ≥33 ie/L were considered positive. The reference ranges were 0.4 to 4.8 × 10 −3 IU/L for TSH and 12–22 pmol/L for fT4. At study site 2 (Regional Hospital Horsens, Denmark), TPOab, TGab, s-progesterone, and fT4 were measured by electrochemiluminescence immunoassays (Cobas 8000 Roche Diagnostics). For fT4, the detection level was 1.3 pmol/L with CV max at 14 and 37.2 pmol/L of 6 and 8%, respectively. For s-progesterone, the detection level was 0.7 nmol/L with CV max at 37 nmol/L of 8%. For TPOab, the detection level was 10 × 10 3 ie/L with CV max at 46 ×10 3 and 109 ×10 3 ie/L of 16 and 10%, respectively. For TGab, the detection level was 25 ×10 3 ie/L with CV max at 95.7 × 10 3 and 336 ×10 3 ie/L of 20 and 12%, respectively. TSH was analyzed by chemiluminometric method using Atellica IM analyzer (Siemens Healthcare Diagnostics Inc., USA) with a detection level of 0.01 × 10 −3 ie/L, CV max at 0.688 × 10 −3 and 29.61 ×10 −3 ie/L of 12 and 10%, respectively. At study site 2, a TPOab value ≥33 ie/L and a TGab level ≥436 ie/L were considered positive. The reference range was 0.3 to 4.5 × 10 −3 IU/L for TSH and 12–20 pmol/L for fT4. Data are presented as mean with standard deviation (SD), median with interquartile range (IQR), or number ( n ) and frequencies (%). The effect of progesterone LPS on thyroid function (TSH and fT4) was analyzed using linear mixed modeling with time and randomization group (±progesterone LPS) and their interaction as fixed factors and participants as random effects. The unequal standard errors and correlations in the two groups were taken into account in the analysis when relevant. Model validation was performed by inspection of QQ-plots of the residuals. We investigated the difference in change from hCG trigger (day 0) to the day of pregnancy testing (day 16). In prespecified analysis, we tested whether the effect of progesterone showed interaction with thyroid autoimmunity (TAI; TPOab and/or TGab positive) and conception by including the relevant interaction term in the model, e.g. TAI status × study group and pregnancy × study group, and in an exploratory analysis, TPOab × study group. We also examined the effect of blastocyst transfer on day 6 versus transfer on day 7 by estimating the interaction of transfer day and development in TSH. Furthermore, we explored whether the effect of progesterone differed according to a high TSH (above the 50th percentile) versus a low TSH level (below the 50th percentile) on trigger day by adding the interaction term of randomization and trigger-day TSH level to the primary linear mixed model described. The association between s-progesterone and TSH was explored using linear regression adjusted for age, BMI, and TAI. We applied an intention-to-treat approach and included all randomized participants in the analysis. The statistical significance level was set at 0.05. The included sample size allowed the detection of a between-group difference in TSH of 0.48, assuming a standard deviation of the TSH difference of 0.83, α 0.05, and β 0.80. Analyses were performed using STATA version 15 (STATA Corp. LP, USA). All women enrolled in the trial provided written informed consent for trial participation, including a separate informed consent for participation in the sub-study and storage of blood samples in a biobank. The study received ethical approval across Denmark by the Capital Region of Denmark Centre for Ethics (H-18025839) and the Danish Medicines Agency (63569). Study medication was provided to the patients free of charge.

Of the 92 eligible women, 85 women fulfilled the inclusion criteria and none of the exclusion criteria and were included in the study ( Fig. 1 ). Forty-five women were randomized to receive progesterone LPS and 40 women were randomized to no LPS. All women received the allocated intervention; one was lost to follow-up. Overall, randomization was successful with comparable groups ( Table 1 ). The participants were 34 ± 4 years of age and had a BMI of 23 (IQR: 21; 26), and 25% were positive for thyroid autoimmunity (TAI). The frequency of TPOab differed between groups, however, with TPOab positivity of 2% vs 23%, P = 0.005, whereas the frequency of TGab was comparable, P = 0.97 ( Table 1 ). Sixty-six percent of women had previously conceived, while 22% had previously given birth. The most common cause of infertility in the study population was male factor infertility. Reproductive outcomes are given in Table 2 . In the progesterone LPS group, 44% conceived compared with 55% in the no LPS group, P = 0.33. Flowchart. LPS, luteal phase support. Baseline characteristics of study groups. Data are presented as n (%), mean ± SD, or median (IQR). Hypertension, thyroid disease, diabetes type 1 or 2, asthma, inflammatory bowel disease, treatment with anticoagulants, depression or anxiety, endometriosis, allergy or atopic dermatitis, or others. Participants might have more than one cause of infertility. Female factor includes tubal factor and endometriosis. Defined as either TPOab positive or TGab positive. Progesterone levels were unmeasurable in 55 participants. BMI, body mass index; IVF, in vitro fertilization; ICSI, intracytoplasmic sperm injection; FET, frozen embryo transfer; TSH, thyrotropin; TPOab, thyroid peroxidase antibody; TGab, thyroglobulin antibody; AMH, anti-Müllerian hormone; LH, lutropin; FSH, follicle-stimulating hormone. Reproductive outcomes in women randomized to luteal support with vaginal progesterone (LPS) or no LPS. Data are presented as n (%). Pregnancy test at day 16 was missing in two women (one from each randomization group), none of whom were registered with pregnancy at a later time point. Confirmed with fetal heartbeat using ultrasonography at gestational week 8. Positive hCG at day 16 but no vital fetus at the transvaginal ultrasound scan at gestational week 8. Loss of pregnancy after verification of a clinical pregnancy at gestational week 8. hCG, human chorionic gonadotropin. In response to progesterone LPS, s-progesterone increased ( Fig. 2 ). However, TSH and fT4 level trajectories were not significantly affected by progesterone LPS ( Fig. 3A and B ). Furthermore, we did not observe any interaction with thyroid autoimmunity, neither when testing for interaction of progesterone LPS and TAI (TGab + TPOab) ( Fig. 4 ) nor when testing for interaction in the presence of TPOab ( Fig. 5 ). Furthermore, the development in TSH did not differ with the day of blastocyst transfer on either day 6 or 7(Supplementary Figure 1 (see section on Supplementary materials given at the end of the article)). Plasma progesterone on five specific time points from the day of hCG trigger (day 0) until day 16 in women receiving progesterone luteal phase support or no LPS. LPS, luteal phase support. Progesterone LPS was initiated in the ‘LPS group’ from day 4 after hCG trigger and is indicated with an arrow. Effect of progesterone luteal phase support on thyroid function. TSH levels (A) or free T4 levels (B) in women receiving progesterone luteal phase support (LPS) or no LPS measured at five specific time points. Linear mixed modeling was used to estimate significance levels. The number of participants was 77 (day 6), 80 (days 11), 78 (day 14), and 81 (day 16). TSH, thyrotropin; T4, free thyroxine; LPS, luteal phase support. Progesterone was administered as vaginal suppositories (100 mg thrice daily) starting on day 4 (marked with an arrow) following ovulation trigger with human chorionic gonadotropin. Effect of progesterone luteal phase support on thyroid function and the interaction with thyroid autoimmunity (TGab and/or TPOab). The impact of TAI on (A) TSH trajectory and (B) free T4 levels during luteal phase was tested by including the interaction term of TAI and study group (LPS) in the linear mixed model. The P -values refer to the significance level of the interaction term. The arrow marks the initiation of progesterone LPS in the ‘LPS group’ from day 4 after ovulation trigger. TGab, thyroglobulin antibody; TPOab, thyroid peroxidase antibody; TSH, thyrotropin; T4, free thyroxine; TAI, thyroid autoimmunity (TGab and/or TPOab); LPS, luteal phase support. Effect of progesterone luteal phase support on thyroid function and the interaction with TPOab. Trajectories of (A) TSH and (B) free T4 levels during luteal phase in response to progesterone. Linear mixed modeling was used to estimate significance levels. The impact of TPOab was tested by including the interaction term of TPOab and study group (LPS) in the model. P -values refer to the significance level of the interaction term. The arrow marks the initiation of progesterone LPS in the ‘LPS group’ from day 4 after ovulation trigger. TPOab, thyroid peroxidase antibody; TSH, thyrotropin; T4, free thyroxine; LPS, luteal phase support. The luteal phase TSH trajectories, in response to progesterone LPS, were independent of whether the TSH level was high on the hCG trigger day (Supplementary Figure 2A), or whether pregnancy was achieved (Supplementary Figure 2B). We found no association between s-progesterone and TSH at hCG trigger day ( β = 0.12; 95% confidence interval: −0.02; 0.27; P = 0.087).

Treatment duration, treatment dose, and formulation of progesterone supplements vary between the few available studies and, therefore, contribute to differences in findings. Furthermore, a longer observation period might have given different results as we only investigated thyroid function through the luteal phase until day 16 after hCG trigger. However, the neutral effect of progesterone LPS on thyroid function within the short-time window examined in our study is important as it demonstrates that with a normal thyroid function, progesterone does not affect the thyroid function, and thus, thyroid testing is not required in the time period until after a pregnancy is confirmed. The possible coactivator-mediated interaction between receptors for progesterone and THs is complex ( 18 ). It is unknown whether the effect of progesterone varies with the TSH level. In our study, we did not detect deviation in results when comparing women with a low-normal or high-normal TSH level. However, all women had a normal thyroid function, and our finding is not readily translatable to situations with an abnormal TSH level. The strength of our study is the randomized design and the frequent thyroid testing. Overall, the presence of TAI (TGab and TPOab) was as expected ( 21 ); however, the frequency of isolated TPOab was lower and differed among randomization groups. We speculate that this may reflect recommendation of previous guidelines to offer levothyroxine for euthyroid TPOab-positive infertile women who would then be non-eligible for this study ( 22 ). We performed interaction analysis and found that the occurrence of TPOab did not change the overall results of a neutral effect of progesterone LPS on TSH and fT4. This is the largest RCT to examine the impact of natural progesterone LPS on thyroid function during mNC-FET treatment. We demonstrate that progesterone does not affect TH level trajectories in euthyroid women in the earliest period during embryo implantation, indicating no need for additional thyroid function testing until pregnancy has been confirmed. Future studies should examine possible long-term effects of progesterone LPS on thyroid function.

In this study, randomized allocation to either progesterone LPS or no LPS did not affect the TSH or fT4 level trajectories during the luteal phase. The presence of thyroid autoimmunity did not change these findings. Only a few studies examined thyroid function in response to treatment with progesterone without estrogen ( 14 ). One RCT ( n = 69) investigating the effects of progesterone in postmenopausal women found an increase in the fT4 level and a trend toward a decrease in the TSH level after 12 weeks of treatment ( 10 ). A small RCT ( n = 8) in a similar population of postmenopausal women found a decrease in the nocturnal TSH level in response to three weeks of non-micronized progesterone treatment ( 9 ). A non-randomized comparative study of women treated with 150 mg depot medroxyprogesterone acetate showed slightly increased fT4 levels but similar TSH levels after 12 months compared to women treated with a copper intrauterine device ( 15 ). The authors suggested that the mechanism might be a suppressed activity of the deiodinases responsible for conversion of T4 to triiodothyronine (T3), resulting in a compensatory increase in fT4 to maintain T3 levels. In contrast, no increase in fT4 was observed during short-term treatment with vaginally applied natural progesterone in our study. Progesterone does not affect the TH-binding protein, such as estrogen ( 7 ). However, progesterone has been suggested to affect the production of THs and the impact of THs at the transcriptional level or by interacting with the TH receptor. The presence of progesterone receptors in human thyroid follicular cells has been detected in some ( 16 ), but not all studies ( 17 ). In a study of cultured normal human thyroid cells, treatment with progesterone significantly enhanced the upregulatory effect of high-dose TSH treatment on several genes important for thyroid function and growth, including an upregulation of the enzyme thyroperoxidase involved in TH production ( 8 ). The addition of mifepristone, a progesterone antagonist, abolished this effect. These data suggest that progesterone affects the production of THs, but whether the effect is dependent on the TSH level is not clear. Other in vitro cell studies have demonstrated a complex coactivator-mediated interaction between the progesterone and the TH receptors ( 18 ). Thus, the effect of THs on target tissue may be modulated by progesterone, as shown in the human placenta ( 5 ). Ovarian stimulation affects the level of TSH, although data on the degree and duration have been diverging. A systematic review found that the evidence was inconclusive, but that TSH concentrations may be increased during or within 1 month after OS ( 19 ). In the very recent ATA guideline on thyroid disease in preconception, pregnancy, and postpartum, it was recommended to postpone a TSH measurement until two weeks after ovarian stimulation ( 20 ). With the postponed FET in our study, it seems unlikely that thyroid function tests were affected by prior ovarian stimulation.

Thyroid hormones (THs) are vital to reproductive health and pregnancy maintenance ( 1 ). The level of THs is meticulously controlled by negative feedback from the pituitary, reflected in the thyrotropin (TSH) level. Any inadequacy of THs, either changes in production or changes in the free hormone fraction, will thus read out in the level of TSH. Another hormone important for pregnancy maintenance is progesterone ( pro gestation ). In early pregnancy, progesterone is produced by the corpus luteum and supports the endometrium, preparing it for implantation ( 2 ). Thus, progesterone is essential for development and receptiveness of the endometrium, and low levels of progesterone are associated with miscarriages ( 3 ). The progesterone receptor is a steroid receptor, and it belongs, like the thyroid hormone receptor (THR), to the nuclear receptor superfamily. They exhibit striking structural and functional similarities ( 4 ). Progesterone affects the expression of several THR isoforms on the protein and mRNA level in human villous trophoblasts ( 5 ). Progesterone supplementation in the luteal phase of the natural frozen embryo transfer cycle has been the standard of care without any real proof of concept ( 6 ). Whether treatment with progesterone affects thyroid function in a fertility setting is not clear. Only a few studies examined the thyroid function in relation to endogenous or exogenous progesterone. A previous study showed that the administration of synthetic progesterone (medroxyprogesterone acetate) does not alter the level of thyroxin-binding protein (TBG) ( 7 ). However, in vitro cell studies ( 8 ) suggest effects of natural progesterone on TH production, and two small randomized controlled trials (RCTs) found that progesterone seems to increase the free thyroxin (fT4) level and decrease the TSH level ( 9 , 10 ). Therefore, we aimed to test whether luteal phase support (LPS) with natural progesterone affects the thyroid function as assessed by TSH and fT4.

DGL, AVG, EE, and UBK have no conflicts of interest. MS reports an unrestricted research grant paid to institution from Gedeon Richter to fund the present work, a travel grant from Gedeon Richter paid to institution, and personal honoraria for lectures from Gedeon Richter and IBSA. KL reports a research grant paid to institution from Gedeon Richter, personal consulting fees from Ferring Pharmaceuticals, personal honorarium for lecture from Ferring Pharmaceuticals and Merck, and personal travel grants from Ferring Pharmaceuticals and Merck. AP reports research grants from Gedeon Richter, Ferring Pharmaceuticals, and Merck; personal consulting fees from Gedeon Richter, Ferring Pharmaceuticals, and Merck; personal honoraria for lectures from Gedeon Richter, Ferring Pharmaceuticals, Merck, and Abbott; and a travel grant from Gedeon Richter. All other authors declare no competing interests.

De-identified individual data, including the data dictionary defining each variable in our dataset, and the code used to analyze the data will be available to researchers upon reasonable request if the research perspectives do not compete with our own plans of publication. We ask that all data-sharing requests be mailed to the lead investigator, AP.

AP and KL designed the main study, with contribution from MS. UBK and EE conceived the sub-study. AP was responsible for funding acquisition. MS was the daily primary coordinator of the study and the data curator. UBK was the study coordinator at Horsens Hospital. MS, UBK, AVG, KL, and AP participated in the inclusion of patients and collection of data. DGL analyzed the sub-study data and wrote the first draft of this paper. AP and KL have accessed and verified the data. All authors approved the decision to submit for publication, participated in thorough revision, and approved the final manuscript.