Discussion
Main Findings
When the pregnancy began, there was a significantly higher rate of late pregnancy loss among the PCOS population, without an increase in the rate of early clinical pregnancy loss. However, in the final adjusted model, PCOS was no longer associated with increased risk of late pregnancy loss. Adverse pregnancy outcome in PCOS may be influenced by increased BMI and underlying medical conditions rather than an independent effect of PCOS.
Strengths and limitations
There are some limitations to acknowledge. First, due to the retrospective character, information about preexisting conditions and gestation-related health issues could be incomplete and underreporting. Insulin concentrations were measured in only a few of the PCOS cases, we did not explore the correlation between the pregnancy insulin resistance and the occurrence of GDM. Secondary, the lines between spontaneous and induced abortion was not clearly distinguished. However, the proportion of induced abortion in this study is negligible, because women are conceived by ART thus artificial abortion is performed very seldom, and only on maternal medical indication. Finally, as we restricted our analyses to pregnancies conceived by ART, our results may not be generalizable to women with natural pregnancies. To draw firm conclusions on the risk of pregnancy loss in PCOS cases, multi- center studies cross country is needed.
The major strength of our study is in its real-world based data with a large sample size of women. Unlike most previous studies, we were able to distinguish between early and late pregnancy loss. Evaluating risk of pregnancy loss at different gestational ages is critical in understanding its etiology and in counseling pregnant women about their possibility of pregnancy loss. Additionally, we also calculated the risk for pregnancy loss stratified by plurality of the pregnancy sac on early ultrasound, which strengthens our findings.
Interpretation
While previous studies have shown that women with PCOS are more prone to suffer from early pregnancy loss,19-22 we show here, as Sterling et al.23 suggested, that risk of early pregnancy loss did not differ markedly between PCOS cases and controls. Furthermore, findings in the current study extended upon previous literature. PCOS cases in the previous study were within the normal range of BMI which does not therefore represent the whole spectrum of PCOS. They studied pregnancy outcomes after fresh embryo transfer only, whereas pregnancy losses in subsequent frozen embryo transfer were also included in our study.
An important finding of the present study was that late pregnancy loss appeared to have stronger associations with PCOS than early pregnancy loss. After 13 weeks of gestation, women with PCOS have been shown to be at higher risk of pregnancy loss, regardless the plurality of the pregnancy. In the adjusted model 1, late pregnancy loss was associated with the diagnosis of PCOS. The retrospective nature of our study design makes it difficult to elucidate the reasons for this finding. This may reflect the hypothesis of a different etiology of pregnancy loss in first and second trimester.
Some authors have argued that the risk of pregnancy loss is more related to the elevated BMI which is known to be associated with an increased risk other than PCOS status.24-26 This is in congruence with the findings of this study. Although the cause of this association between PCOS and obesity remains unknown, overweight is present in 30%~50%.27-29 and in the present study 33.7% had BMI ≥25 kg/m2 and 9.2% with BMI ≥30 kg/m2. We noticed that the potential negative impact of PCOS was eliminated once BMI were taken into account in the fully adjusted Model 2. In line with our results, Joham et al,3 suggested that PCOS was not independently predicts higher risk of a pregnancy loss. However, as overweight and obesity often coexisting with PCOS,27-30 it is debatable whether data should be controlled for BMI.
Significant differences were found in maternal preexisting medical conditions with a markedly increased risks of hypertension and diabetes were noted in women with PCOS compare with women with non-PCOS, which was supported by other researches.30-33 A recent review of PCOS patients, derived from a UK general practitioner research database with a mean age of 27 years followed for a median period of 4.7 years, demonstrated that women with PCOS had a higher systolic blood pressure.30 Reports suggest that a woman with PCOS may have a fourfold increased risk of developing diabetes, and a 33% risk of impaired glucose tolerance.33 Though overweight seems to be the most important predictor, the effect of comorbidities also remained statistically significant after multivariable analysis (Table 4 model 2)
Women with PCOS are prone to undergo ART, with its higher frequency of twins and multiple pregnancy. The loss of pregnancy due to multiple pregnancies have been evaluated in PCOS patients. Mikola et al.34 found that the higher incidence of poor obstetric outcomes of PCOS pregnancies could partly explained by the increased number multiple pregnancies. As double-embryo transfer is still common, we here, subdivided the pregnancy loss rate according to the numbers of gestational sac in early ultrasound. Our results here do not suggest that the number of embryos transferred or multiple pregnancies alone increases the risk the pregnancy loss among patients with PCOS (Table 2 and Table 3). But the results should not be taken as a plea for DET or twin pregnancies for those with PCOS. On the contrary, higher rate of late pregnancy loss was observed when DET was performed in both of the two groups, confirming that SET is a logical practice.35 In line with previous studies,36 we found that 25.8% percent of multi gestational pregnancies that progressed to a livebirth delivery experienced loss of at least one fetus during the pregnancy. More worryingly, spontaneous reductions in IVF/ICSI twin pregnancies have been suggested to be a possible cause of the increased morbidity in IVF singletons.36-38 Message above is critical for the whole cohort of infertility patients since they frequently ask for more embryos to be transferred to secure a maximum chance of success.
In the present study we found women with PCOS were slightly younger than women without the diagnosis. PCOS women frequently exhibit menstrual irregularities such as oligomenorrhea, hence are more likely to require medical assistance. With the awareness of the potentially reduced fertility, they could have started trying to conceive earlier. This hypothesis is supported by an observational study, which reported that women with PCOS are also more likely to have had their first pregnancy at a younger age.33 Advanced maternal age is strongly correlated with early pregnancy loss in the study (Table 3), although we adjusted for maternal age in the multivariate analysis, a residual effect could still be possible. However, there was no association with maternal age and late pregnancy loss in PCOS or non-PCOS cases, before and after multivariable analysis (Table 4). These findings need to be confirmed in future studies.
Various studies suggested that females with PCOS who conceive might suffer from pregnancy-related complications such as gestational diabetes,39 pregnancy induced hypertension39,40 to a higher extent in comparison to controls. Varies studies have shown that infants born to women with PCOS are also predisposed to many adverse health outcomes.2,23,40 We have consistently reported an increase in maternal and neonatal complications for women with PCOS, even analysis was restricted to singletons. But we failed to find a correlation between PCOS and risk of caesarean section, which does not correspond with the findings of other studies. Differences can probably be explained by the high incidence of caesarean section in China, either caused by social or clinical factors, particular when women undergoing ART.