The first striking observation of this study is that women with PCOS display very high levels of total testosterone in their follicular fluid as compared to controls (44 times) and to previously reported circulating plasma levels (27). Although a difference was expected, its magnitude is surprising considering that plasma testosterone levels at the time of oocyte retrieval, after ovarian stimulation, were not reported to be as high in women with vs without PCOS: they were increased only 1.3-fold in a previously published IVF study (28). This difference may be explained by the fact that the follicular fluid matches closely the intra-ovarian milieu where androgens are directly produced. Moreover, as it has been previously shown that follicular fluid testosterone levels in unstimulated follicles show a 2-fold increase in women with PCOS when compared to women without (27), the difference in our study may be due to the gonadotropin stimulation prior to oocyte retrieval. It has been reported that PCOS ovaries are more sensitive to gonadotropin stimulation than non-PCOS ovaries (29).
To our knowledge, this study is the first to demonstrate that the ovarian environment of PCOS women is characterized by higher lipid concentrations, independently of the degree of obesity. Triglycerides were the most important contributors to this difference and may come from the circulation or local NEFA converted to triglycerides by surrounding cells. These results support the hypothesis that PCOS ovaries are over-exposed to lipids that are likely to induce cellular lipotoxicity, at least during the ovarian stimulation of the IVF procedure. When looking at plasma lipid concentrations in women with PCOS (outside IVF), three studies showed higher levels (30–32), while others found no difference (33–35). Only one study assessed NEFA levels in the follicular fluid, during IVF stimulation, and showed no difference compared to non-PCOS (19). This may be due to the lack of power of the limited sample size (N = 6 women with PCOS) or lack of hyperandrogenicity in PCOS, as androgen concentrations were not reported (19). More importantly, they did not assess triglyceride levels, which were the most elevated fat in our study and can readily provide NEFA to cells and cause lipotoxicity.
Using the same sample as in this study (16), we have previously shown that follicular fluid lipid levels correlated with follicular testosterone levels (16). It is therefore likely that the higher ovarian exposure to lipids found in women with PCOS may contribute to their increased androgen production (36). Several mechanisms have been proposed to explain lipotoxicity (37), including the accumulation of toxic lipid metabolites such as diacylglycerol and ceramides (38, 39), which are produced when the cell’s ability to metabolize NEFA is exceeded and/or defective. Although our study was not designed to provide insight into the presence of these mediators in follicular fluid, we found indicators of inefficient mitochondrial fatty acid oxidation in women with PCOS. In our study, triglycerides and acylcarnitines were increased twofold or more in the follicular fluid of women with vs without PCOS, which was significant even after correction for BMI. This was also the case when comparing BMI-matched groups. This confirms that the excess of lipids found in the intra-ovarian milieu of women with PCOS are taken up and metabolized by surrounding cells to a greater extent than in non-PCOS. Triglycerides are converted to NEFA that can enter the mitochondria for beta-oxidation only after transformation in acylcarnitines with longer chains of carbons, such as palmytoylcarnitine, oleoylcarnitine and linoleoylcarnitine (C16) (40). Once inside the mitochondria, they are beta-oxidized to produce acetylcarnitine that will exit the mitochondria (C3). Hence, a higher ratio of C16/C3-acylcarnitnies is a good marker for ineffective mitochondrial NEFA metabolism (28). We found that the C16/C3 ratio was 35% higher in women with vs without PCOS, which was still significant after correction for BMI and this difference remained when comparing BMI-matched groups, although it was no longer statistically significant with this smaller sample size. Thus, during ovarian stimulation, mitochondrial lipid metabolism is impaired in PCOS ovaries, favouring the accumulation of toxic lipid metabolites and lipotoxicity. This defect could be due to either lipid overload and/or mitochondrial dysfunction, which is supported by the fact that ovaries from high-fat diet mice tend to accumulate lipids and their oocytes exhibit features of mitochondrial dysfunction (18).
Elevated IL-6 levels are also characteristic of the PCOS ovarian environment during IVF treatment. TNF-α levels, on the other hand, were not increased in this group as compared to the non-PCOS group. It is however possible that TNF-α may not be able to enter the follicular fluid as easily as IL-6. Higher plasma concentrations of IL-6 (41) or macrophage release of TNF-α under hyperglycemic conditions (34) have been observed in women with PCOS, who also show increased macrophage infiltration in their ovaries (42). The origin of inflammatory markers in follicular fluid is unclear, as it they could originate from the systemic circulation, be produced by ovarian cells, or released by macrophages infiltrating ovaries. In addition, systemic or intra-ovarian inflammation could contribute to the excess of androgen production in women with PCOS (16) and NEFA may directly induce inflammation through activation of toll-like receptors (43), the inflammasome, and the ERK/MERK pathway in women with PCOS (44). Inflammation may therefore have contributed to PCOS hyperandrogenism in our study, but cannot explain our results alone, because differences in follicular fluid levels of testosterone, lipids or lipid metabolites between groups were independent of both BMI and IL-6. On the other hand, higher follicular fluid levels of IL-6 in PCOS women were no longer significant after correction for BMI, and the difference was reduced and not statistically significant using BMI-matched paired analysis. These analyses strongly suggest that the increased inflammatory markers found in the follicular fluid of PCOS women are due to an obesity-related mechanism. However, further studies are needed to assess the role of intra-ovarian inflammation in the development of PCOS.
When comparing the IVF clinical outcomes, no difference was detected between the two groups, despite a non-significant twofold increase in pregnancy rate in women without PCOS. Consistent with our results, a meta-analysis reported that women with PCOS undergoing IVF had more oocytes retrieved per procedure and a lower fertilization rate than control women, but similar pregnancy rates (45). Fatty acid composition has been shown to affect distinct features of IVF, such as oocyte quality and number (46), which may explain the lower fertilization rate and a trend towards lower pregnancy rates in women with PCOS.
This study as some limitations. First, the small sample size of women with a definite diagnosis of PCOS, which reduces statistical power. Accordingly, differences that are not statistically significant should be considered with caution. However, differences that are statistically significant are robust and can be conclusive. Second, the technique of egg retrieval may have caused bleeding during sampling and samples were frozen before centrifugation, which may have resulted in cell lysis. It is therefore possible that some samples were contaminated with blood or intra-cellular content. Since the objective of this study was to use the follicular fluid as a surrogate of intra-ovarian milieu, the contamination by other intra-ovarian fluid is still relevant to the results. Furthermore, we believe that such contamination was probably minimal. Of note, follicular fluid testosterone levels found in our study cannot be compared with previously published as testosterone was measured by radioimmunoassay (47–50), whereas here the gold standard LC-MS/MS was used, which is more sensitive and specific. Third, ovarian stimulation and the use of metformin may have affected some parameters, such as lipids and testosterone. Indeed, gonadotropin stimulation during the IVF protocol may have increased testosterone more in women with PCOS than in non-PCOS (29), while metformin may have lowered lipid levels in women with PCOS. Since a higher proportion of women with PCOS were under treatment with metformin, the difference in follicular fluid lipid content between groups may be even higher than observed.