Our study has showed that the systematic measurement of peak serum E2 levels in COS-IUI cycles does not reduce the risk of multiple pregnancies when strict cancelation criteria based on the patient’s age and follicular monitoring are applied. We found that peak serum E2 was not predictive of the risk of MP with an area under ROC curve of 0.60 (0.52–0.69), and that the correlation with the number of follicles ≥ 10 mm and ≥ 14 mm was moderate (linear correlation r of 0.43 and 0.41, respectively.
In the past two decades, several studies have tried to identify the risk factors associated with twin and higher order multiple pregnancies (HOMP) in COS-IUI cycles, and develop prediction models that would allow to lower the MP rate without decreasing the overall success rates. Among these risk factors, serum E2 level has been found to be linked to MP rates, but only at very high levels and when combined with an excessive follicular response. In an analysis of 441 pregnancies following COS-IUI, out of which 9% were MP, Gleicher et al.10 found a significantly higher risk of MP when peak serum E2 level was > 1385 pg/mL or when there were > 6 pre-ovulatory follicles on US10. Tur et al.12 reported a 15.6% twin pregnancy and a 5.7% HOMP rate in 1878 pregnancies following COS-IUI. The HOMP rate was 19% when peak serum E2 was > 862 pg/ml with > 5 follicles > 10 mm in women ≤ 32 years of age12. The same authors reported in a later study that the use of a prediction model that includes the woman’s age, the number of preovulatory follicles, and the peak serum E2 level led to 285% reduction in the rate of HOMP20. By applying the same predictive model in our algorithm (Fig. 1), we found a twin pregnancy rate of 7.7% and a HOMP rate of 0.4%, both considerably lower than those cited. However, it should be noted that COS with 150 IU or higher of exogenous gonadotropins is considered a risk factor of MP when compared to COS with 50 or 75 IU13,16. The mean total FSH dose used by Tur et al. was 1120 IU compared to 744 IU in our study, and the mean number of follicles ≥ 10 mm was 4.7 compared to 2.3 in our study12. The use of lower gonadotropin doses and a strict predictive model allowed us to lower the MP rate while maintaining an acceptable overall clinical pregnancy (13.4%) and live birth rate (10.8%).
We aimed to assess whether the systematic measurement of peak serum E2 level, as per our algorithm, played a part in lowering the MP and HOMP rates. We found that serum E2 level was not a predictive factor of MP, with an area under ROC curve of 0.60 (0.52–0.69) (Fig. 3), and an aOR of 1.06 (0.85–1.32) (Table 4). Moreover, the Pearson coefficient showed only a moderate correlation between serum E2 level and the number of follicles ≥ 10 mm (r = 0.43) and ≥ 14 mm (r = 0.41). On the other hand, multivariate analysis also found that the treatment duration, the gonadotropin doses used, and the number of follicles ≥ 10 mm and ≥ 14 mm were not predictive factors of MP (Table 4 and supplementary data). This could be the consequence of our adherence to strict cancelation criteria and the use of relatively low gonadotropins doses (50 to 100 IU) which limited the number of growing follicles. Indeed, there were only 3 cases (0.08%) where the cancelation criteria were not respected.
There were 9 cases (0.2%) that were canceled because the peak serum E2 level was very high (> 900 pg/mL), despite a normal ultrasound with no sign of excessive response to COS. Our rate is markedly lower than the 5.5% (68/1327) reported by Tur et al. in COS-IUI cycles12. Interestingly, the rate of HOMP in that study was 8% in women > 32 years of age and 12% in women ≤ 32 years of age12. In this instance, the measurement of peak serum E2 levels might have been useful, and prevented potential MP by signaling an excessive response to COS that was not suspected on US monitoring. However, the occurrence of these cases is low (0.2%).
Based on our findings, we decided to create a new algorithm for cycle cancelation that is more adapted to our current practice in COS-IUI cycles (Fig. 4). There are two major changes in the new algorithm: first, we modified the threshold for peak serum E2, and increased it from 862 pg/mL to 1000 pg/mL. The initial threshold was based on the studies by Tur et al.12,20, who used radioimmunoassay (RIA) to measure serum E2 levels, while at our center, we use chemiluminescent immunoassay (CLIA). In general, the difference between these two methods is as follows: CLIA = 1.04 (RIA) + 20 pg/mL21. The calculated new threshold stands at 920 pg/mL, but we rounded it up to 1000 pg/mL to make it easier to use in daily practice. The second major modification is to abandon the systematic measurement of peak serum E2 levels, and only proceed with it in specific cases, depending on the US monitoring results. The new indications for serum E2 measurements are: 3 follicles ≥ 14 mm in women ≤ 32 years of age, and 4 follicles ≥ 14 mm in women > 32 years of age (Fig. 4).
The main limitation of our study is the retrospective design. It would be interesting to prospectively compare the two cancelation policies, one with and one without measurement of peak serum E2 levels. The main strength of our analysis in the inclusion of a large cohort of 3630 COS-IUI cycles over a period of 9 years. Moreover, and to the best of our knowledge, this is the first study assessing the role of serum E2 measurement in COS-IUI cycles when a strict cancelation policy is applied. The results of our analysis could be helpful for clinicians in their everyday practice.
In conclusion, our study has shown that, in COS-IUI cycles, when strict cancelation criteria based on the woman’s age and the number of growing follicles are used, the systematic measurement of peak serum E2 levels does not help reduce the rate of MP. Serum E2 levels could still help clinicians in the decision-making process in certain equivocal and problematic cases, when associated with the other criteria. In the near future, artificial intelligence models that incorporate all the risk factors of MP (woman’s age, number of growing follicles, type, duration and dose of ovarian stimulation, as well as serum E2 levels in certain cases) could calculate the risk of twin and HOMP in any given clinical situation, and help guide physicians and couples in deciding whether to proceed or cancel the treatment cycle.