In routine practice, the starting dose of gonadotropins is individualized to assure optimum safety and efficiency of the controlled ovarian hyperstimulation (COH) in order to obtain sufficient ovarian response and reduce the risk of developing OHSS. Thus, if no previous cycles have been performed, the choice of the starting dose of gonadotropins will be based on a prediction of the ovarian response, which is built based on some patient characteristics like patient’s age, ovarian reserve, day 3 FSH, and antral follicle count (AFC) . In addition, the Ovarian Sensitivity Index (OSI), which is a marker link between the number of retrieved oocytes and the total administered dose of FSH, has been introduced recently to estimate ovarian sensitivity to exogenous gonadotropins, and its values negatively correlated with age and positively with AFC and the circulating levels of AMH . PCOS women have higher antral follicular counts and higher levels of AMH and estradiol, which exaggerates their response and sensibility to COH [41, 42] and explains the higher OSI values of this population. However, that puts PCOS women at increased risk to develop OHSS [43–45], so they are usually stimulated with a lower starting dose and require a lower total dose of gonadotropins during COH. In the current study, we noted that stimulating PCOS women with the GnRH agonist long protocol led to a significantly higher number of retrieved oocytes compared to the controls, and this increase in the oocytes number covered both the mature and immature oocytes. In addition, the number of fertilized oocytes and obtained embryos trend to be significantly higher in the PCOSA group compared to the ControlA group. Interestingly, similar effects could not be detected between PCOS and controls during the GnRH antagonist protocol. Our results on the long protocol were consistent with several previous clinical studies [24, 46, 47]. However, they partially disagreed with the results of the Arabzadeh et al. study , which reported insignificant differences in the number of retrieved oocytes, maturation rate, fertilization rate, high-quality embryos rate, and implantation rate between PCOS and controls undergone the long agonist protocol. Nevertheless, Arabzadeh et al. study  infertility inclusion criteria for all cases included unexplained infertility, infertility due to sperm or tubal abnormalities, and endometriosis. Therefore, including women with endometriosis might have influenced the final results since endometriosis has a negative impact on IVF outcomes, and it is associated with lower oocyte yield, lower implantation rates, and lower pregnancy rates . On the other side, our results on the GnRH antagonist protocol were consistent with the Afiat et al. study , which could not detect any significant differences in the number of MII oocytes and MI oocytes between PCOS and controls that treated with GnRH antagonist protocol. Differently, the study of Le et al.  showed higher numbers of retrieved oocytes and mature oocytes in the PCOS group compared to the control one during the GnRH antagonist protocol. Similarly, Nikbakht et al.  also found a higher number of retrieved oocytes in the PCOS group. Indeed, focusing only on the number of obtained oocytes without taking into account the doses that were used for stimulation would lead to misleading results. Unfortunately, many of these studies did not mention the doses that were provided to PCOS or control women. In our study, women in the PCOSAnta group were stimulated with (1779.55 ± 702.87) IUs and provided (17.73 ± 9.76) oocytes and (10.18 ± 5.55) MII oocytes while the ControlAnta were stimulated with (2468.18 ± 879.53) IUs and provided (16.24 ± 8.99) oocytes and (9.03 ± 5.34) MII oocytes. On the other hand, during Le et al. study , women in the PCOSAnta group were stimulated with (1820.69 ± 332.06) IUs and provided (18.85 ± 9.41) oocytes and (14.97 ± 7.43) MII oocytes while in the ControlAnta were stimulated with (2005.60 ± 379.69) IUs and provided (11.48 ± 5.51) oocytes and (9.51 ± 4.7) MII oocytes. However, Nikbakht et al.  did report the total stimulation doses that were used in the studied groups. Although the women of the PCOSAnta group of our study and Le et al. study  were stimulated with similar doses, women in the ControlAnta group of our study were stimulated with a little higher dosage of gonadotropins compared to those from Le et al. study  and produced higher number of retrieved oocytes. Thus, the increase in stimulators dose did not arise from a lower response to gonadotropins. Therefore, we think that stimulating the Control‑Anta group in our study with a little higher dose might have prevented the differences between the two groups (PCOS‑Anta vs Control‑Anta) from reaching the significance level and that PCOS women would respond more aggressively to COH irrespective of the protocol used. That also can be confirmed by the fact that OSI values in our study differ significantly between PCOS women and controls independently from the protocol used. Thus, we encourage using the OSI index in clinical studies to remove the confounding effects of using different doses of gonadotropins.
Several studies raised some concerns regarding the oocyte quality of PCOS women. However, the available data are still conflicted. Niu et al.  suggested an association between abnormal lipid metabolism and oocyte competence, and they concluded that the high concentrations of linoleic acid and palmitoleic acid both in the plasma and in the follicular fluid of obese PCOS women might contribute to the poor pregnancy results of IVF in this population. In addition, Lai et al.  reported that the increased reactive oxygen species (ROS) expression levels in PCOS granulosa cells greatly induced cell apoptosis, which further affected the oocyte quality and reduced the pregnancy results. Based on our results, there were no differences in maturation rate, fertilization rate, highquality embryos rate, cleavage rate, implantation rate, oocytes morphology or clinical IVF/ICSI outcomes between the PCOS and the control women either during the GnRH agonist protocol or the GnRH antagonist one. This partially agrees with the prospective study of Sigala et al. , which also could not find any differences in the oocyte morphology, maturation rate, fertilization rate, or highquality embryos rate between women with polycystic ovarian morphology and women with normal ovarian morphology. However, they reported a higher implantation rate, ongoing pregnancy rate, and delivery rate in the PCO group compared to the control one. It should be taken into account that Sigala et al.  included both women with PCOS and women with only PCO in the PCO arm and the participants were stimulated with a combination of the GnRH agonist and the GnRH antagonist protocols. In addition, the authors declared that they did not exclude low responder patients from the control group, which may explain the better clinical outcomes in the PCO group. Similarly, Afiat et al.  reported comparable oocyte nuclear maturity and embryo grades between PCOS and non-PCOS women during the GnRH antagonist protocol. However, time-lapse studies on embryos development ended up with contradicted results [5, 53–55]. The study of Chappell et al.  showed that embryos from PCOS women displayed a faster growth rate at t7, t8, and t9 compared to controls, while those from hyperandrogenic PCOS showed a faster growth rate at t5, t6, t7, t8, t9, and morula stage. Similarly, Sundvall et al.  reported a shorter time to initiate compaction and reach the morula stage; and a shorter duration of the fourth cleavage division in the PCOS embryos compared with the non‑PCOS ones, but the kinetic at other time-points were similar. On the other hand, Le et al.’s study  found no differences in morphokinetics or incidence of abnormalities between PCOS and non-PCOS embryos. However, the percentage of t2 stages which fell in the “optimal range” (> 24 h and < 28 h) was significantly lower in the PCOS group than in the control group. On the contrary, Wissing et al.  reported a significant delay in time to two pronuclei breakdown, first cleavage, and cleavage to three, four, and seven cells in embryos from hyperandrogenic PCOS compared to controls. It is worth mentioning that the assessment of embryo development was carried out for a shorter duration in the Le et al.  (for 48 hours or to the six-cell stage), and Wissing et al.  (to the eight-cell stage in most cases as the embryos were transferred on Day 2, and only the remaining embryos were studied until Day 5 or Day 6) studies. In addition, the studies differ in the protocol of stimulation as it was the long GnRH agonist in the Wissing et al.’s study , the GnRH antagonist in Le et al.’s study , and a combination of the two in Sundvall et al.’s study . However, none of these studies could detect any differences in implantation rate, pregnancy rate, or live birth rate between PCOS and non-PCOS women, which suggests that although PCOS exaggerates ovarian response to stimulation, it does not have a detrimental impact on oocytes quality or competence. This also agrees with the results of the retrospective study of Vas et al., which reported similar rates of fertilization, implantation, and clinical pregnancy from the oocytes that were taken from PCOS donors and non-PCOS donors , and even if PCOS led to some minimal deviations in embryo developmental process, these deviations might not be clinically important.
In the current study, FF PLGF levels were comparable between PCOS and controls during both; the long GnRH agonist protocol and the flexible GnRH antagonist one. In addition, FF PlGF levels were negatively correlated with age and total gonadotropins dose and positively correlated with OSI in the PCOSAnta, ControlA, and ControlAnta groups, but not in the PCOSA group. Moreover, FF PlGF levels were positively correlated with the number of MII oocytes in the PCOS‑Anta group and the number of retrieved oocytes in the Control‑A group. Based on our recent work, the long GnRH agonist protocol is associated with significantly higher levels of FF PlGF compared to the flexible GnRH antagonist one both; in PCOS and normo-ovulatory women [27, 28]. Therefore, the more aggressive stimulation effects of the long agonist protocol on the PCOS women might have disturbed the correlation between the FF PlGF levels and the OSI values and/or the total gonadotropins doses in the PCOSA group. Our results partially agree with the results of the study of Nejabati et al. , which could not detect any significant differences in FF PlGF levels between poor responders, normoresponders, and high responders among nonPCOS women that undergone IVF/ICSI cycles with the long GnRH agonist protocol. However, they showed that FF PlGF levels were significantly and negatively correlated with age and total FSH dose, but not with the number of retrieved oocytes or the OSI. On the other hand, PlGF/sFlt1 ratios were significantly and negatively correlated with age and fertilization rate while positively correlated with the number of retrieved oocytes, the number of obtained embryos, and the OSI. These differences in correlations might be related to the fact that the correlations in the Nejabati et al.  study were assessed among the total number of participants independent of the response classification. In addition, although there were not any differences in FF PlGF levels among various responder groups, PlGF/sFlt-1 ratios differ significantly between poor responders and high responders. Differently, Tal et al. reported significantly higher FF PlGF levels and lower FF sFlt1 levels in PCOS women compared to controls. They also demonstrated that FF PlGF levels positively correlated with the number of oocytes and the serum levels of AMH while negatively correlated with age. However, in the Tal et al. study , the women were simulated using both the GnRH agonist and GnRH antagonist protocols, and the correlations were evaluated among the total number of participants, i.e. they included PCOSA, PCOSAnta, ControlA, and ControlAnta. Regarding the differences in FF PlGF levels between PCOS and controls, although they found significantly higher levels of FF PlGF in PCOS women, while we could not detect any differences between the two populations, we do not think our results disagree with theirs. As we previously mentioned, PlGF levels differ significantly between the flexible GnRH antagonist protocol and the long GnRH agonist one, both in PCOS and normo-ovulatory women [27, 28]. However, even if we compared the PCOS groups (PCOSA + PCOSAnta) together with the control groups (ControlsA + ControlsAnta), we could not detect any significant differences in FF PlGF levels between PCOS and controls women (data not shown). Nevertheless, the total gonadotropins doses were significantly different between (PCOSA vs ControlA), (PCOSAnta vs ControlAnta), and (PCOSA + PCOSAnta vs ControlA + ControlAnta) in our study, but not in the Tal et al. one . Thus, in our opinion, it is all related to the consumed dose of gonadotropins and the OSI of the population. Since PCOS women have higher OSI and are usually considered higher responders to gonadotropins compared to controls, stimulating them with similar gonadotropins doses will produce more oocytes and require higher levels of PlGF to accomplish that response taken into account that PlGF controls ovarian angiogenesis and follicular development [14, 15, 57]. In addition, we do not think that this effect is specific to PCOS subjects, but to all high responders as Nejabati et al.  also could not detect any significant differences in FF PlGF levels between poor responders, normoresponders, and high responders when the provided gonadotropins doses were significantly lower between high responders vs poor responders and high-responders vs normoresponders.
Based on our results, FF PlGF were comparable between pregnant and nonpregnant women, both in PCOS and normoovulatory women, independently of the protocol used, which also had been confirmed by the ROC curve analysis. That agrees with the results of Nejabati et al.  on nonPCOS women during the long agonist protocol. Although PlGF levels are positively correlated with the OSI, which reflect the ovarian response to stimulation, other factors like the degree of male infertility, sperm/oocytes genetics integrity, infertility duration, and endometrial receptivity may also influence the pregnancy achievement.
Strengths, limitations, and future research:
To the best of our knowledge, this is the first study that investigated the correlations between the FF PlGF levels and the IVF/ICSI outcomes in PCOS and normo‑ovulatory women and the dependency of these correlations on the COH protocol used. In addition, it is the first study to examine in detail the impact of PCOS on the oocyte morphology during both the long GnRH agonist protocol and the flexible GnRH antagonist one. However, our study has some limitations. First, due to the limited budget, our study was only concerned about the total FF PlGF levels and not the levels of the free form of PlGF (PlGF/sFlt‑1 ratio). In addition, our study only included normo‑ovulatory and PCOS women, so further research is needed to clarify whether similar correlations would be noted between FF PlGF levels and IVF/ICSI outcomes from other populations with different ovarian responses, e.g. aged women, poor responders, or endometriotic women.