Low progesterone levels on the day before natural cycle frozen embryo transfer are negatively associated with live birth rates

STUDY QUESTION: Are progesterone (P) levels on the day before natural cycle frozen embryo transfer (NC-FET) associated with live birth rate (LBR)? SUMMARY ANSWER: Regular ovulatory women undergoing NC-FET with serum P levels <10 ng/ml on the day before blastocyst transfer have a significantly lower LBR than those with serum P levels >10 ng/ml. WHAT IS KNOWN ALREADY: The importance of serum P levels around the time of embryo transfer in patients undergoing FET under artificial endometrial preparation has been well established. However, no study has analyzed the importance of serum P levels in patients undergoing FET under a true natural endometrial preparation cycle. STUDY DESIGN, SIZE, DURATION: This was a retrospective cohort study including 294 frozen blastocyst transfers under natural cycle endometrial preparation at a university-affiliated fertility centre between January 2016 and January 2019. PARTICIPANTS/MATERIALS, SETTING, METHODS: All patients had regular menstrual cycles and underwent NC-FET with their own oocytes. Only patients who had undergone serum P measurement between 8 am and 11 am on the day before FET were included. Patients did not receive any external medication for endometrial preparation or luteal phase support. Patients were divided into two groups according to serum P levels below or above 10 ng/ml on the day before FET. Univariate analysis was carried out to describe and compare the cycle characteristics with reproductive outcomes. To evaluate the effect of P, a multivariable logistic model was fitted for each outcome after adjusting for confounding variables. MAIN RESULTS AND THE ROLE OF CHANCE: Mean serum P levels on the day before FET were significantly higher in patients who had a live birth compared to those who did not (14.5 +/- 7.0 vs 12.0 +/- 6.6 ng/ml, 95% CI [0.83; 4.12]). The overall clinical pregnancy rate (CPR) and LBR were 42.9% and 35.4%, respectively. Patients in the higher P group (>10 ng/ml) had a higher LBR (41.1% vs 25.7%: risk difference (RD) 15.4%, 95% CI [5; 26]) and CPR (48.6% vs 33.0%: RD 15.6%, 95% CI [4; 27]). Patients with higher serum P levels on the day before FET (63% of patients) had an improved LBR (odds ratio: 1.05; 95% CI [1.02; 1.09]). Women with serum P levels <10 ng/ml on the day before FET (37% of patients) had significantly higher weights (62.5 +/- 9.9 vs 58.1 +/- 7.1 kg, 95% CI [1.92; 6.90]) and BMI (22.9 +/- 3.6 vs 21.6 +/- 2.7 kg/m(2), 95% CI [0.42; 2.25]) compared to patients with P levels >10 ng/ml. LIMITATIONS, REASONS FOR CAUTION: The main limitation of our study is its retrospective design. Other potential limitations are the detection of LH surge through urine testing and the inclusion of patients who did and did not undergo preimplantation genetic testing for aneuploidies. The protocol used in our institution for monitoring NC-FET does not look for the onset of progesterone secretion by the corpus luteum, and a slow luteinisation process or delay of corpus luteum function cannot be ruled out. WIDER IMPLICATIONS OF THE FINDINGS: We provide evidence that a minimum serum P threshold (P >10 ng/ml) might be required for improved reproductive outcomes in NC-FET. This result suggests that there are different mechanisms by which P is produced and/or distributed by each patient. This study also provides an excellent model to evaluate the impact of luteal phase defect through NC-FET. A prospective evaluation to assess whether P supplementation should be individualised according to patient's needs is necessary to support our findings.