The characteristics of PCOS, such as androgen excess, obesity, insulin resistance, and abnormal metabolism, may increase the risk of obstetric and neonatal complications . Many studies have shown that PCOS is related to adverse perinatal outcomes, such as gestational diabetes mellitus , premature rupture of membranes , and pregnancy-induced hypertension . In addition, PCOS in pregnancy is also associated with adverse neonatal outcomes, and an increase in the incidences of premature delivery, intrauterine growth restriction, abnormal birth weight (smaller or greater than gestational age infants), and admission to the neonatal intensive care unit . After adjusting for confounding factors in the logistic regression analysis, we identified BMI, menstrual pattern, endometrial thickness on the day of hCG administration, and the number of eggs harvested among PCOS patients who received IVF/ICSI-assisted pregnancy, as well as the incidences of abortions, premature births, and pregnancy complications, as independent risk factors for adverse pregnancy outcomes.
Among patients with PCOS, excessive androgen levels interfere with normal follicular development and result in the stagnation of follicular development or even atresia. Excessive androgen levels typically lead to oligomenorrhea or amenorrhea , and excessive levels of testosterone or its metabolites can also cause insulin resistance in female adipocytes through androgen receptors . This further reduces the secretion of insulin-sensitive fat factor and adiponectin, and leads to fat accumulation and obesity . In our study, the BMI of patients with amenorrhea was higher than that of patients with oligomenorrhea, and BMI was an independent risk factor for adverse pregnancy outcomes. A meta-analysis of 40 observations and investigations indicated that gestational diabetes mellitus occurred in PCOS patients with migraines with aura. The increased incidence of adverse perinatal outcomes, such as eclampsia and preterm birth is related to BMI , consistent with the results of our study. However, some researchers believe that perinatal complications, such as gestational diabetes mellitus and gestational hypertension in PCOS patients, are affected by PCOS itself and have nothing to do with BMI [20, 21]. To sum up, as an intuitive clinical observation index, BMI may be used as a predictor of adverse pregnancy outcomes among PCOS patients, but not in isolation.
Currently, the influence of different PCOS phenotypes on the occurrence of adverse pregnancy outcomes remains controversial . A recent meta-analysis showed that the cumulative pregnancy rate of the PCO + HA + oligoovulation group was lower than that of the PCO + oligoovulation group . A prospective cohort study showed that PCOS patients with a hyaluronic acid + menstrual disorder + PCO phenotype had a higher risk of adverse maternal and infant outcomes than those without the menstrual disorder phenotype . A study of patients with PCOS with normal or rare menstruation and natural pregnancies showed that the risk of pregnancy complications in patients with PCOS may be affected by menstrual patterns, while the risk of pregnancy complications in patients with normal ovulation was similar to that of the normal population; however, no analysis was conducted among patients with PCOS who had amenorrhea . The focus of this study was to compare patients with PCOS that had amenorrhea and those with oligomenorrhea. The results showed that the incidence of adverse pregnancy outcomes (abortions, premature births, and pregnancy complications) and abnormal newborn weight (VLBW) was higher in patients with amenorrhea than in patients with oligomenorrhea. The rates of adverse outcomes in the amenorrhea group were high, suggesting that the menstrual pattern of patients with PCOS affects maternal and fetal outcomes. Among patients with PCOS, the increased risk of pregnancy complications and abnormal birth weight may be due to placental damage during pregnancy . Patients with PCOS who have abnormal menstruation, particularly those with amenorrhea, should receive more attention in terms of perinatal monitoring. Their management and follow-up should also be strengthened to avoid complications, such as gestational diabetes mellitus and abnormal weight of newborns.
The increased AMH production by granulosa cells in patients with PCOS can inhibit the expression of aromatase RNA in granulosa cells, reduce the transformation of androgens to estrogens, and increase the level of androgens in follicles . In our study, patients with amenorrhea had higher AMH levels compared to those with oligomenorrhea. Studies have shown that patients with PCOS who have elevated AMH levels have more obvious endocrine and lipid metabolism disorders, suggesting that AMH can predict the severity of insulin resistance and HA . Another retrospective study divided patients with PCOS into groups according to menstrual patterns (regular menstruation, hypomenorrhea, and amenorrhea). There were significant differences in 18F-fluorodeoxyglucose uptake, homeostatic model assessment of insulin resistance results, and triglyceride levels among the three groups, suggesting that menstrual patterns can be used as proxies of glucose and lipid levels among patients with PCOS. They can also be used as the simplest and most direct indicator of abnormal metabolism . Metformin may improve endometriosis by upregulating the expression of glutamate transporter 4 in the endometrium of patients with PCOS . Other studies have shown that the level of 20-hydroxyeicosatetraenoic acid (HETE), a metabolite of arachidonic acid, is positively correlated with obesity and insulin resistance , and abnormal HETE levels may lead to gestational diabetes mellitus  and gestational hypertension . Therefore, in patients with amenorrhea, more severe metabolic disorders may cause abnormal elevations in HETE levels, leading to complications during pregnancy.
According to the National Institutes of Health standard or Rotterdam standard—a retrospective study of patients with PCOS who have different phenotypes and normal control groups  compared patients with the (ANOV + HA or PCO) phenotype to those with the (PCO + ANOV or HA) phenotype—compared to the normal control group, the uterine volume and endometrial thickness of the type group were smaller, and this indicated that different phenotypes of PCOS patients affected the uterine volume and endometrial thickness . The results showed that after adjusting for confounding factors, the endometrial thickness on the day of hCG administration was an important factor that affected the perinatal outcome and neonatal birth weight (odds ratio, 0.89; 95% confidence interval, 0.80–0.99) . Intrauterine studies of poor outcomes such as thin membrane thickness leading to fetal growth restriction, premature delivery, and an increased incidence of small for gestational age infants showed that the occurrence of these phenomena may be related to the decline in trophoblast cell invasion and abnormal spiral arterial remodeling .
In the present study, we found that the number of eggs harvested was an independent risk factor for adverse pregnancy outcomes. A retrospective analysis of 65,868 single live births showed that harvesting too many eggs during IVF can lead to premature birth and LBW, which may be due to abnormal ovarian function or changes in the endometrial environment caused by the high levels of physiological hormones . Endometrial dysfunction and abnormal trophoblast invasion and implantation may cause abortions and pregnancy complications in patients with PCOS .
In this study, the relationship between menstrual patterns among patients with PCOS and IVF/ICSI-assisted pregnancy outcomes were analyzed for the first time. It was proven that menstrual patterns were an independent risk factor for adverse pregnancy outcomes (including abortions, premature births, and pregnancy complications). Furthermore, to reduce the impact of different items on pregnancy outcomes, we only included patients who received the long-acting long-term follicular phase.
Simultaneously, our study had some limitations. We performed only a retrospective study, which did not consider all confounding factors. Most newborn data were collected through telephone follow-up rather than direct access to medical records, which might have led to a lower incidence of neonatal abnormalities than the actual incidence. Thus, large sample, multicenter, prospective studies are still needed to confirm our findings.