A higher incidence of smooth endoplasmic reticulum clusters with aromatase inhibitors

Abstract Purpose This study aimed to analyze whether a regimen of aromatase inhibitor (AI) could reduce the occurrence of smooth endoplasmic reticulum clusters (sERCs) in oocytes. Method(s) The AI and the clomiphene citrate (CC) regimens were compared, regarding the sERC (+) rates and the serum estradiol and progesterone levels on the date of hCG administration, and the duration of AI, CC, and hMG administration. Result(s) The occurrence of sERCs in oocytes from patients treated with AI was significantly higher than that in oocytes from those treated with CC. Both the serum estradiol and progesterone levels were found to be significantly higher in sERC (+) than in sERC (‐) cycles. With regard to the CC cycles, no significant differences were detected. The duration of AI and hMG administration was longer for sERC (+) than for sERC (‐) cycles. Conclusion As AI did not reduce the occurrence of sERCs, the elevation of estradiol may not be the cause of sERC occurrence but a consequence. Considering the higher levels of progesterone and longer duration of hMG in sERC (+) cycles, the negative effects of premature luteinization, which frequently occur with the AI protocol, should be investigated further.


| INTRODUC TI ON
Smooth endoplasmic reticulum clusters (sERCs) have been known to appear occasionally in human oocytes. A cumulative number of abnormalities in live births, following the transfer of embryos from sERC (+) cycles and oocytes, has been reported, [1][2][3][4][5] although healthy babies can be born from embryos derived from sERC (+) oocytes. 6,7 The clinical impact of sERC-positive oocytes remains controversial; however, two recent studies reported that sERC-positive oocytes had negative biological impact. Canto et al reported that sERC-positive oocytes were prone to a higher frequency of aberrant spindle formation. Furthermore, sERC-positive oocytes have been strongly associated with aberrant actin distribution in their sub-oolemmal regions. 8 Otsuki et al reported that the incidence of meiotic and mitotic cytokinesis failure was higher in embryos derived from sERCpositive oocytes than in embryos derived from sERC-negative oocytes. 9 An embryonic cell that experiences cytokinesis failure during meiosis results in 3PN, which during mitosis could become tetraploid and might cause abnormal chromosomal configurations in the embryo. 9 However, the mechanism underlying sERC formation is still unknown. A high concentration of estradiol per oocyte when ovulation is triggered may be an indicator of sERC predisposition, 1 as it is well known that estradiol levels increase in proportion to the growth of ovarian follicles. It has also been reported that sERC formation is related to the higher serum estradiol 1,2,10 and progesterone 11 levels detected in sERC-positive cycles.
The enzyme aromatase synthesizes estrogen, and aromatase inhibitor (AI) suppresses the production of estrogen. 12 Ovarian stimulation with AI has been proposed as one of the treatments for infertility. 13 AI blocks the aromatization of androgen to estrogen, and this has a similar effect to that of clomiphene citrate (CC), producing more follicles. Although the level of estrogen will decrease, AI does not have the same antiestrogen effects as CC. With this hypothesis in mind, this study aimed to retrospectively analyze whether a regimen of AI, to stimulate the growth of ovarian follicles, could reduce the occurrence of sERCs in oocytes.

| MATERIAL S AND ME THODS
This was a retrospective cohort study of patients whose serum anti- The clinical outcomes of patients undergoing frozen-thawed single embryo transfers prior to the end of December 2017 were analyzed.

| Stimulation protocols
The ovarian stimulation protocols were chosen depending on the patient's age and AMH level. Patients treated with AI and CC were in the same classification, and the choice of AI or CC was according to the patients' preferences; therefore, the two regimens, AI and CC, are comparable. With the CC protocol, 50 mg/day of clomiphene citrate (Clomid, Fuji Pharma, Tokyo, Japan) was adminis-

| Timing of ICSI, detection of sERCs, and the cryopreservation of embryos
ICSI was performed on MII stage oocytes 4 hours after oocyte retrieval. MI stage oocytes were used for ICSI only when they reached MII stage maturity within 8 hours of oocyte retrieval. This approach was a consequence of reported data which demonstrated that there is no significant difference between the outcomes of MII oocytes and that of MI-MII oocytes which reach MII on the day of oocyte retrieval.
Oocytes were classified morphologically into sERC-positive and sERC-negative oocytes at the time of ICSI under inverted microscopes (OLYMPUS, Tokyo, Japan at X400 magnification). The presence of sERCs was defined as being of a size sufficient to be observable by microscope and was generally more than 10µm in diameter. Sperm were prepared using a density-gradient centrifugation

| Statistical analysis
Statistical analysis was performed using the χ 2 -test or Student's t test where appropriate, employing JMP Software, Version 11 (SAS Institute Japan). Multiple logistic regression analysis was also used to obtain odds ratios in the presence of several explanatory variables. Differences were considered statistically significant when the P-value was < .05.  Table S1 shows the occurrence of sERCs in oocytes from patients treated with AI and CC, and compares the differences in patients when they were divided into two age categories (40> and 40≦). There were no statistical differences between the two regimens regarding age, AMH level, and the number of ICSI cycles. The occurrence of sERCs in oocytes from patients treated with AI was also significantly higher than that in oocytes from those treated with CC in each age category.
With regard to CC cycles, serum estradiol and progesterone levels on the date of hCG were 1024.2 ± 425.0 pg/mL and 0.82 ± 0.49 ng/ mL in sERC (+) and 909.9 ± 444.7 pg/mL and 0.80 ± 0.47 ng/mL in sERC (−) cycles. As compared to the significant differences in serum estradiol and progesterone levels found in AI cycles, no significant differences were detected in CC cycles (E2: P = .304; P4: P = .826) (  Regarding the high-quality blastocyst formation rate, no statistical difference was discovered for either the AI or CC regimens (AI: 5.0%  Table 2)). When the patients were separated into two age categories (40 >and 40≦), no significant difference was found between the AI and CC regimens among patients who were 40 or older. In contrast, among patients who were younger than 40, the duration of AI or CC administration and hMG administration was significantly longer for the AI regimen than CC (Table S3). Between sERC (+) and sERC (−) negative cycles for the AI and CC regimens, no significant differences were found regarding implantation, miscarriage or live birth rates or the occurrence rate of congenital abnormalities (Table S4).
No significant differences in clinical outcomes were observed for either regimen when the patients were divided into two categories regarding their age (40> and 40≦) (Table S5). Please note, embryos derived from sERC (+) oocytes were only transferred when no embryos derived from sERC (−) oocytes were available for transfer.

| D ISCUSS I ON
The initial goal of this study was to reduce the occurrence of sERCs with the application of AI regimen, which has been reported to lower E2 levels. However, contrary to expectation, it was found that the occurrence of sERCs was significantly higher in oocytes from patients treated with AI than in those treated with CC. Furthermore, an increased occurrence of sERCs remained constant for the AI regimen when patients were divided into two age categories. This substantiates the hypothesis that the occurrence of sERCs is directly congruent with an elevation in progesterone. Considering the higher levels of progesterone in sERC (+) cycles, it may also be related to prolonged ovulation, with the negative effect of premature luteinization on oocytes. This may be explained by a previous report 14 that the occurrence of sERCs from AI with a GnRH antagonist regimen was no higher than that in other ovarian stimulation regimens, as administration of a GnRH antagonist suppresses premature P4 elevation.
As premature luteinization frequently occurs with the process of AI without GnRH antagonist protocols, further study will be required TA B L E 2 Duration and total dosage of administration of stimulation drugs, serum estradiol and progesterone levels, and embryo development rates for each regimen in sERC (+) and sERC (−) cycles among patients treated with AI and CC to elucidate the causes of the occurrence of sERCs. It may also be beneficial to study whether sERC occurrence rate is higher in random-start controlled ovarian stimulation when it is started at luteal phase with elevated progesterone. As the results of this study come from patients whose AMH level was less than or equal to 1.0 ng/mL, the occurrence of sERCs among patients with the presence of higher AMH levels is still unknown.
A further point of critical importance: the 3PN formation rate among embryos derived from sERC (+) oocytes was significantly higher than the 3PN formation rate of sERC (−) oocytes. This was consistent with previous findings, that the rate of 3PN formation is higher in sERC (+) than sERC (−) oocytes. 9 As a failure of the second polar body extrusion results in 3PN formation, and mitotic failure has also been reported to be higher in embryos derived from sERC (+) oocytes, the transfer of embryos derived from sERC (+) oocytes should be performed with caution. Furthermore, a recent paper found that the presence of sERCs in oocytes alters the molecular structure. This has the potential to cause mitotic and meiotic errors and negatively affect the spindle assembly, organization of cytoskeleton, the microtubules, and the mitochondrial structure and activity. 15 sERCs usually appear in MII stage oocytes rather than in MI and GV stage oocytes and also appear when the metaphase stage is extended. 9 It has also been suggested that the occurrence of sERCs could also be a sign of prolonged cytoplasmic maturation prior to the triggering of LH surges in controlled ovarian stimulation cycles. 1 This study also found that the duration of AI and hMG administration was longer for sERC (+) than for sERC (−) cycles. Additionally, the duration of administration was longer for AI than for CC regimens. Thus, the occurrence of sERCs could be unrelated to estradiol levels. As the half-life of letrozole is shorter than that of CC, 16 the influence of letrozole should be less significant than that of CC.
There was no significant difference in clinical outcomes between transferred embryos derived from sERC (+) or sERC (−) oocytes.
Moreover, no significant difference was detected between the AI or CC regimens, even when patients were separated into two age categories (40> and 40≦).
In this study, comparison was limited to the two aforementioned ovarian stimulation protocols. Other stimulation protocols, such as long, short, and antagonist protocols, could not be compared with AI in this study, as the patients' backgrounds (age and levels of AMH) varied too greatly. However, our results provide an important insight for future studies investigating the mechanism underlying sERC formation.

ACK N OWLED G EM ENT
We would like to thank Dr Michio Yamamoto, Associate Professor at Graduate School of Environmental and Life Science, Okayama University for his valuable advice on statistical analysis.

CO N FLI C T S O F I NTE R E S T
Hiroe Saito, Junko Otsuki, Hiromi Takahashi, Rei Hirata, Toshihiro Habara and Nobuyoshi Hayashi declare that they have no conflicts of interest.

H U M A N/A N I M A L R I G HTS
This article does not contain any experimental studies with human