Adiposity and lipid metabolism indicators mediate the adverse effect of glucose metabolism indicators on oogenesis and embryogenesis in PCOS women undergoing IVF/ICSI cycles

Polycystic ovary syndrome (PCOS), characterized by oligo-anovulation, hyperandrogenism, and polycystic ovarian morphology, is the most common endocrine disease affecting 4 to 21% of reproductive-aged women [1‐3]. PCOS has diverse lifelong clinical manifestations of reproductive, metabolic, and psychological disorders, such as anovulatory infertility, dyslipidemia, obesity, impaired glucose tolerance (IGT), Type 2 diabetes mellitus (T2DM), insulin resistance (IR), cardiovascular disease, low self-esteem, psychosexual dysfunction, depression, and anxiety [1, 2, 4, 5]. PCOS women undergoing adverse assisted reproductive technology (ART) are at high risk of adverse IVF outcomes [6]. The pathogenesis of PCOS is complex, including genetic and epigenetic susceptibility, hypothalamic and ovarian dysfunction, hyperandrogenism, hyperinsulinemia, and adiposity-related mechanisms [2, 7].

As common manifestations of PCOS, maternal lipid and glucose metabolism dysregulations have been suggested to contribute to abnormal embryonic development [8‐12]. In vitro fertilization (IVF)-based studies indicated that poor embryogenesis and pregnancy outcomes in diabetic and obese mice result from the maternal effects on the oocytes rather than the uterine environment, suggesting an oocyte origin of adverse reproductive outcomes [11, 13]. IVF/intracytoplasmic sperm injection (ICSI) provides an excellent means to study the effects of maternal factors on oogenesis and embryogenesis. The recent clinical studies have shown that dyslipidemia, obesity, glucose homeostasis dysregulation, IR, and metabolic syndrome (MetS) contribute to ART outcomes in PCOS women. A multicenter randomized controlled trial of 1508 PCOS women observed that MetS affects female fecundity in forms of longer infertility duration, poorer ovarian response, fewer retrieved oocytes, fewer available embryos, and lower chance of cumulative live birth [14]. Several studies have also reported the impact of sole obesity and IR on pregnancy outcomes in PCOS women undergoing IVF/ICSI [15‐17]. However, these results remain controversial [18‐20], and it is not easy to explore causal paths due to the close interactions among metabolite profiles.

The causal roles of body mass index (BMI) and IR in PCOS etiology have been proven via Mendelian randomization (MR) analyses [21‐23]. However, the causal relationship between glucose and lipid metabolism in PCOS women remains uncertain. A small MR analysis in the Netherlands indicated that dyslipidemia (i.e., high plasma triglycerides (TGs) levels) seem to be responsible for hyperglycemia and T2DM risk, and large (L) and extra-large (XL) high-density lipoprotein (HDL) lipid components might protect against increasing fasting glucose level [24]. However, a recent MR study based on the genome-wide association summary statistics of Europeans reported bidirectional causality among serum glycemic and lipid traits [25]. Therefore, we assume the potential mediation role of adiposity and lipid metabolism indicators on the associations between glucose metabolism indicators and IVF/ICSI outcomes in PCOS women.

When compared with MR analysis, mediation analysis is another valuable tool for exploring causality by testing whether a variable is a mediator and what proportion of the role it plays [26]. In this study, we aimed to determine whether adiposity and lipid metabolism indicators mediate the effect of glucose metabolism indicators on early reproductive and pregnancy outcomes in PCOS women undergoing IVF/ICSI cycles.

In this study focused on women with PCOS, we observed significant dose-dependent correlations between glucose metabolism indicators and IVF/ICSI early reproductive outcomes and between glucose metabolism indicators and adiposity and lipid metabolism indicators. Moreover, we observed significant dose-dependent correlations between adiposity and lipid metabolism indicators and IVF/ICSI early reproductive outcomes. Our estimates suggested that serum TG, serum HDL-C, serum LDL-C, and BMI mediate the inverse associations between glucose metabolism indicators and all six IVF/ICSI early reproductive outcomes in PCOS women. We also observed that serum TC mediates the inverse associations between glucose metabolism indicators and high-quality embryo count in PCOS women.

Glucose homeostasis dysregulation, IR, obesity, and dyslipidemia are critical clinical manifestations of PCOS. The metabolisms of glucose and lipid are often intertwined and regulated [34, 35]. Most evidence suggests altered lipid concentrations are secondary to IR, IGT, or diabetes [36]. High-sugar diet-induced diabetes and genetically diabetes (db/db) mice can cause obesity and dyslipidemia by either microbiota imbalance or deficiency of leptin receptor [37, 38]. Obesity and dyslipidemia could also contribute to the pathogenesis of IR or glucose homeostasis dysregulation. Evidence from animal studies has proved that obesity and dyslipidemia can cause hyperinsulinemia and IR in high-fat diet (HFD)-induced obese and genetically obese (ob/ob) mice [39]. Lipids can influence glucose metabolism via several pathways, including disturbing insulin secretion from pancreatic β-cells [40], apoptosis [41], stress responses (e.g., oxidative stress, endoplasmic reticulum stress, and autophagy) [42], and mitochondrial dysfunction [43]. As an instrumental variable method investigating the causal effects of traits using genetic predictors, MR has been used to determine the causal paths among glucose, insulin, and lipid metabolism markers in humans [24, 25, 44, 45]. The causal roles of adiposity and lipid metabolism markers for diabetes have been elucidated [24]. Several studies found that elevated BMI, TG, and small HDL particle levels might increase the risk of diabetes, while elevated large HDL particle levels might have a protective effect [24, 46]. Recently, Zhu et al. systematically investigated the associations among serum HDL-C, LDL-C, TG, fasting insulin, and HbA1c in European populations and illustrated the bidirectional causality of lipids and glycemic traits [25]. They indicated that elevated serum fasting insulin level was a decisive causal factor of low serum HDL-C and high serum TG levels, which in turn would further increase serum fasting insulin [25]. We found significant positive effects of glucose metabolism indicators on serum TC, serum TG, serum LDL-C, and BMI and negative effects of glucose metabolism indicators on serum HDL-C. These findings also prove the close relationships between glucose metabolism indicators and adiposity and lipid metabolism indicators.

We found negative associations between adiposity and lipid metabolism indicators and IVF/ICSI early reproductive outcomes in PCOS women. Numerous previously published literature can support our findings. Increasing BMI was found to be associated with poorer ovarian response [47, 48], fewer retrieved oocytes [47‐49], fewer mature oocytes [49], lower clinical pregnancy rate [16, 17, 50], lower live birth rate [16, 51, 52], and higher risk of miscarriage [48, 51] among PCOS women undergoing IVF/ICSI cycles. High serum TC level was found to be a risk factor for the live birth of IVF/ICSI cycles in PCOS women [52, 53]. The mechanism of obesity and dyslipidemia for PCOS women may be intricate. By promoting the mobilization of free fatty acids, IR is a recognized cause of dyslipidemia in PCOS [54]. Obesity can alter the adipokine distribution in adipocytes, thus, resulting in abnormal activation of pro-inflammatory response, accumulated toxic reactive oxygen species, and aggravated oxidative stress [55]. Inflammation and oxidative stress have been illustrated to cause abnormal oogenesis and folliculogenesis [56]. The damage of obesity to oocytes and embryos may also be related to the alteration of epigenetic reprogramming [56]. Animal studies have elucidated that maternal obesity can cause a remarkable reduction of an essential maternal effect protein named Stella (aka Dppa3/Pgc7) in oocytes and may disrupt the establishment of pronuclear epigenetic asymmetry and impair embryo development via accumulation of maternal 5-hydroxymethylcytosine (5-hmc) modifications and DNA lesions [11].

We also revealed inverse associations between PCOS women’s glucose metabolism indicators and IVF/ICSI early reproductive outcomes after controlling for various confounders and adiposity lipid metabolism indicators. Evidence from previous clinical and animal studies supports our results. A secondary analysis of a multicenter randomized trial illustrated that MetS in PCOS women had a negative impact on IVF cycle stimulation characteristics and clinical outcomes, including higher and longer doses of gonadotropin, lower peak estradiol levels, fewer retrieved oocytes, fewer available embryos, and lower oocyte utilization rate [14]. In another clinical trial, elevated serum insulin, HOMA-IR, and homocysteine levels were negatively associated with oocyte numbers and maturation in PCOS women [19]. The mechanism by which glucose metabolism dysfunction leads to abnormal oogenesis and embryogenesis remains unclear. T2DM was found to increase oocyte mtDNA mutations [57]. Studies also found that IR contributed to apoptosis, oxidative stress, impaired mitochondrial function, abnormal spindles, and misaligned chromosomes of oocytes; thus, leading to fertilization disorders and arrest of embryo development [58]. Moreover, preconception hyperglycemia can reduce the expression of Tet3 dioxygenase in the oocytes, which is responsible for 5-methylcytosine (5-mc) oxidation and DNA demethylation in the zygote. This epigenetic alteration disturbed the subsequent DNA methylation reprogramming of embryos [12]. Our mediation analysis showed that TG, TC, HDL-C, LDL-C, and BMI mediate the associations between glucose metabolism indicators and IVF/ICSI early reproductive outcomes, suggesting that obesity and dyslipidemia may mediate the adverse effect of glucose homeostasis dysregulation and IR on oocyte and embryo quality. As we have discussed above, glucose homeostasis dysregulation and IR have been associated with dyslipidemia and may be the causality of dyslipidemia [25], possibly by increasing mobilization of free fatty acids [54]. Animal and human studies also supported the adverse effect of obesity and dyslipidemia on IVF/ICSI outcomes via inflammation, oxidative stress, and epigenetic mechanisms [49, 55, 57]. However, we must cautiously interpret the mediation analysis’s results because solid assumptions are required to interpret our findings’ causality.

To our knowledge, this is the first study that conducted a mediation analysis to assess the potential role of adiposity and lipid metabolism indicators in the effects of glucose metabolism indicators on IVF/ICSI outcomes in PCOS women. Our findings provide insights into the mechanism of abnormal oogenesis and embryogenesis caused by glucose metabolism dysfunction and show the importance of the dynamic equilibrium of glucose and lipid metabolism in oogenesis and embryogenesis in PCOS. However, some limitations of this study warrant consideration. First, given the retrospective nature of this study, we cannot entirely avoid the bias in data collection. However, the bias in data collection can be eliminated to some extent because the data we studied are objective measures with specific definitions. Second, the database did not fully record information on adjunctive therapies, including diet, exercise, or medication (i.e., metformin) before or during IVF treatment. Future well-designed prospective studies should focus on the impact of adjunctive therapies. Third, multiple hypothesis tests were simultaneously performed to assess the associations between glucose metabolism indicators and adiposity and lipid metabolism indicators. Thus, we cannot entirely rule out the possibility that some of our results are found by chance. Fourth, we only included PCOS women from a reproductive center, thus possibly causing selection bias and limiting the extrapolation of our results to the general PCOS population. However, these findings may have significance for those who seek infertility treatment, which accounts for 40 to 72% of PCOS women [2, 59]. Fifth, although we found that abnormal glucose metabolism is associated with PCOS women’s adverse pregnancy outcomes in this study, we found no mediating role of obesity and dyslipidemia in this relationship. Further mechanism studies are still needed.

In conclusion, in the present study on 917 PCOS women undergoing their first IVF/ICSI cycles, we found that glucose metabolism indicators were associated with adiposity and lipid metabolism indicators (i.e., increased TG, TC, LDL-C, and BMI and decreased HDL-C), and adiposity and lipid metabolism indicators were inversely correlated with IVF/ICSI early reproductive outcomes (i.e., reduced retrieved oocyte count, MII oocyte count, normally fertilized zygote count, normally cleaved embryo count, high-quality embryo count, or blastocyst formation count). Mediation analyses suggested that obesity and dyslipidemia may mediate a significant proportion of the adverse effect of glucose homeostasis dysregulation and IR on IVF/ICSI early reproductive outcomes. Our findings show the importance of preconception glucose and lipid management in PCOS women.