From the patient characteristics, we found a significant difference in the median age and AMH level between the two groups. These findings are similar to previous studies. For example, Lauritsen et al. (2014) reported a lower PCOS prevalence with increasing age (5). It might be caused by the decrease of ovarian reserve or antral follicle as a woman age (6).
Higher AMH levels in PCOS patients also has been reported by numerous studies (4, 7–10). It might be caused by the increased synthesis and secretion of AMH by pre-antral and small antral follicles (3). The level of AMH increases with antral follicle count (AFC) at a consistent rate of 0.2 ng/ml per follicle (11). In addition, granulosa cells in follicles of PCOS patients have been shown to produce 75-fold more AMH than normal (12).
From the IVF process, we found that the PCOS group received significantly less gonadotropin dose during COS. However, their number of oocytes retrieved was found to be higher. Moreover, the number of hyper responders in the PCOS was also significantly higher than the control group. Although all the study subjects received antagonist protocol, different doses may be given. Following our clinic’s protocol, protocol and doses were decided based on age, BMI, AFC, and previous response to stimulation, if any.
Indeed, the major problem of gonadotropin stimulation is the increased risk of multiple pregnancy and OHSS. Therefore, low-gonadotropin stimulation has been introduced and lately has been accepted as best practice for PCOS patients, especially for those who have clomiphene citrate (CC)-resistant (13, 14). In the last few years, a more sophisticated method was introduced. The individualized COS (iCOS) warrants identifying high-risk patients through various biomarkers, with AMH and AFC seem to be promising variables (2, 15). Few studies have reported the use of iCOS led to significantly less OHSS occurrence (16, 17).
Our PCOS group showed significantly higher average numbers of retrieved oocytes despite lower gonadotropin doses administered. Various studies have well documented similar findings. Sahu et al. reported that despite a significantly lower total gonadotropin dose, PCOS and PCO-only groups yielded more oocytes compare to controls (18). However, they also found no differences in oocyte maturation, embryo quality, implantation, and pregnancy rate. We also found the same circumstances in our study. Swanton et al. also found significantly more oocytes retrieved in PCOS and PCO-only groups (n = 14.2 and n = 16.2, respectively) compared to control (n = 10.5). They also found significantly more severe OHSS cases in the PCOS and PCO-only groups, similar to ours.
Curiously, when we tried to determine a cut-off value of AMH levels for hyper-response prediction, it resulted in a poor predictive value. It is presumably the result of the heterogenicity of each antral follicle. While AMH production is high, each follicle’s threshold to respond to ovarian stimulation may vary in women with PCOS. Kim et al. analyzed the correlation between AMH-MoM (AMH level multiples of median) and ovarian sensitivity to gonadotropin stimulation. They found that ovarian sensitivity was not correlated to the AMH-MoM value and even tended to decrease with an increasing AMH-MoM level (19). These findings may suggest that the use of basal serum AMH level as a predictor of ovarian response should be carefully applied, considering other factors in the patients.
Due to its retrospective design and a relatively small number of patients, our study resulted in low power. Moreover, various PCOS phenotypes were recruited in the study, which might play a role in PCOS patients’ response to gonadotropin stimulations. More comprehensive factors and data must be included to further investigate the underlying causes associated with OHSS in PCOS patients.