The present study shows highly variable LP P4 levels in healthy regularly cycling women, and furthermore, that the FP E2 level on days 5 and 10 is predictive of a P4 level ≥ 30 nmol/L on days 20 or 25. Importantly, no other covariates, including FSH level cycle day 5, LH levels during the follicular phase, age, weight, AFC and AMH seem to be associated with the LP P4 level.

The 30 nmol/L luteal phase cut-off for P4

Jordan et al. previously conducted a study in a group of 58 regularly cycling women to explore P4 levels on a daily basis during the LP. The cohort was strictly selected for several parameters, i.e., age, a history of regular menstrual cycles, BMI, and LP length > 12 days. Based on that cohort, luteal phase deficiency (LPD) was defined as an AUC of P4 less than 254.4 nmol days/L during the LP [14]. In another cohort study by Del Pozo et al., LPD was defined as an AUC < 340.26 nmol days/L [21]. Based on these AUC cut-offs and observations, the authors concluded that the most accurate method to predict a low AUC of P4 was one single mid-luteal serum P4 level measurement < 31.8 nmol/L, or the sum of three random serum P4 levels during the LP being less than 95.4 nmol/L.

The AUC in the present study was an estimation based on P4 levels every fifth day, showing significantly lower mean AUC in cycles with no LP P4 levels ≥ 30 nmol/L (118 ± 75 nmol x days/L) compared to a mean AUC of 365 ± 151 nmol days/L in cycles with a P4 level ≥ 30 nmol/L on day 20 or 25.

Interestingly, a study exploring salivary samples consisting of a total of 22 regularly cycling women, demonstrated a significant variation in LP P4 levels between individuals and cycles. The participants collected salivary samples daily over a period of six months, with LP P4 levels calculated as the average of 14 days salivary P4 levels before the onset of menses. The authors focussed on the intra-individual and inter-cycle variation and suggested that changes in calorie intake and physical activity could partially explain this phenomenon [12]. No information on calorie intake or activity was available in the present study, and weight was only measured during the first cycle.

Is a 30 nmol/L P4 level sufficient to support a pregnancy?

Importantly, the present study did not include contraceptive cycles. The objective of the study was to demonstrate the frequency of cycles with low P4 levels and, by extension, the risk of insufficient P4 levels during the early stages of pregnancy. A sufficient P4 during the luteal phase was evaluated in HRT-FET cycles, with a level of approximately 32 nmol/l being proposed [17]. In relation to the natural cycle, Hull et al. investigated the minimum serum P4 level necessary for a successful live birth to occur in 212 natural conception cycles. The authors concluded that no pregnancy occurred if P4 levels were below 27 nmol/L or above 50 nmol/L. [11]. Thus, LPD is a physiological phenomenon that affects conception during natural cycle. This is particularly important given the increasing use of natural cycle for frozen embryo transfer [26] in which some individuals undergoing true natural frozen embryo transfer (tNC-FET) will need additional exogenous luteal phase support (LPS). Gaggiotti-Marre stressed the point by reporting a significantly higher live birth rate (LBR) if P4 levels exceeded 32 nmol/L, the day prior to blastocyst transfer in tNC-FET cycles, compared to cycles with lower P4 levels (41.1% versus 25.7%, respectively) [9]. However, it appears that LPS involving exogenous P4 may compensate for the P4 deficiency. Thus, Lawrenz et al. reported that if exogenous P4 was added from the blastocyst transfer day in a tNC-FET cycle, P4 levels measured before initiating P4 treatment on the transfer day did not affect the ongoing pregnancy rate [15].

Finally, a meta-analysis (N = 489) including three randomised controlled trials showed that exogenous P4 LPS in tNC-FET was superior to no LPS in terms of reproductive outcome with a risk ratio of 1.42 [13].

E2 levels during follicular phase as a predictor of CL function

E2 is primarily secreted by the ovary due to the aromatization of testosterone in the granulosa cell during the FP. After ovulation, the granulosa cells transform into luteal cells, which secrete P4. Thus, it can be hypothesised that the E2 secretion during the FP mirrors granulosa cell function and if the E2 is high during the FP, the likelihood of an optimal P4 production during the LP is increased. In the present study 89% (25/28) of cycles with an E2 level ≥ 345 pmol/L on day 10 of the FP had a P4 level ≥ 30 nmol/L on days 20 or 25 compared to 45% if E2 was lower than 345 nmol/L (OR 10.0, 95% CI [2.6; 38.4]).

LH levels as a predictor of CL function

LH plays a crucial role during ovulation, and the LH surge has been described to consist of three phases, an ascending phase, a plateau phase and a declining phase lasting for a total of 48 h [10]. As for the LH peak, Park et al. reported that it could be highly variable in appearance and subsequently different definitions of LH peaks have been suggested [7, 20].

In the present study, lower LH levels on cycle day 10 and 15 correlated with a P4 level ≥ 30 nmol/L on days 20 or 25 compared to higher levels. The paradox may be explained by the fact that most women had their LH peak between the 10th and 15th cycle day and blood sampling in the present study was limited to every fifth day, resulting in small number of LH peaks being detected. Furthermore, during the early LP, P4 is mainly secreted from the large luteal cells (granulosa cells) which are not LH dependent, unlike during the mid- and late LP where the small luteal cells are taking over which are LH and HCG dependent, originating from the theca cells [6, 8, 19].

FSH or AFC

In the present study, neither FSH levels on cycle day 5 nor AFC were predictive of P4 levels on day 20 or 25. However, due to the small number of women in this cohort (n = 6) with AFC < 5 and only 10 cycles (5 patients) in which the FSH level on day 5 was greater than 10 IU/L, we cannot draw any conclusions on whether AFC or FSH are prognostic for P4 levels during the LP.

Strengths and limitation

The strength of the present study is that 25 women had standardised blood sampling during three consecutive cycles and reduces the risk of changes in lifestyle, including weight and exercise habits which is known to influence steroid levels. Furthermore, all participants were lean, healthy, working women with regular cycles. A limitation is the number of participants and the fact that blood samples were drawn only every fifth day. Ovulation was not detected in the present study, however, as the range of cycle lengths was very narrow (25 to 32 days), the risk of not detecting a mid-luteal P4 rise on either cycle day 20 or 25 is suggested to be low.

With the increasing number of FET cycles being performed in the tNC, we revisited the present data collected during 2011–2012. Importantly, the data question whether all true natural cycles result in optimal LP P4 levels and interestingly, we found that the E2 level on cycle day 10 seems to be a predictor of an optimal FP P4 level. Future research needs to focus on developing diagnostic tools to identify the most optimal natural cycle steroid level.

In conclusion, the present study is the first to suggest a significant correlation between FP E2 levels and a well-functioning CL, secreting a sufficient amount of P4 during the LP. Moreover, our findings again highlight the existence of a significant inter-cycle and intra-individual variation in P4 levels in regularly cycling women. With the increasing use of tNC FET, we underline the need for more research on how to monitor the natural cycle in order to individualise the luteal phase support to achieve the highest reproductive outcome.