This retrospective study screened 943 patients who underwent IVF or ICSI treatment between February and June of 2020 in the Centre for Assisted Reproduction of Shanghai First Maternity and Infant Hospital. Patients were excluded for the study if they: (1) were over 45 years old (N=2), (2) did not undergo oocyte retrieval (N=3), (3) did not obtain transferable embryos (N=50), or (4) did not undergo embryo transfer (N=60). The flowchart of participants included in this study is shown in Figure 1 . This study was approved by the Research Ethics Committee of Shanghai First Maternity and Infant Hospital (KS21211).

All included patients had received controlled ovarian stimulation (COS), including long GnRH agonist (GnRH-a) protocol, GnRH antagonist (GnRH-ant) protocol, short GnRH-a protocol, mild stimulation protocol and progestin-primed ovarian stimulation (PPOS) protocol. The details of each protocol were described previously (16). Oocytes were retrieved 34-36 h after HCG injection. The process of semen preparation, conventional IVF/ICSI, embryo culture and assessment were performed as described previously (17). For women who underwent fresh embryo transfer, one or two high-quality embryos were transferred three or five days after oocyte retrieval under transabdominal ultrasound guidance. Luteal phase support was started on the day of oocyte retrieval as described previously (16). Endometrial preparation protocols and luteal phase support for frozen-thawed embryo transfer were described previously (16). Luteal-phase support was maintained in women with a positive hCG test until ten weeks of pregnancy.

All subjects were measured in light clothes without shoes and hats. Body weight was measured by weight scale to the nearest 0.1kg. Height was measured in meters by a rangefinder. A tapeline was used to measure the WC and hip circumference to the nearest 0.1cm. The WC was measured at the narrowest part of the torso, which is mid-way between the inferior border of the rib cage and the superior aspect of the iliac crest. The hip circumference was measured at the level of the trochanters, which is the maximal extension of the buttocks. WHR was calculated by dividing waist circumstance by hip circumference. Weight, height, waist and hip circumference were measured twice by a nurse throughout the study, and the mean value of these measurements was used in our study to minimize errors.

The primary outcome was live birth after the first embryo transfer of the IVF cycles, defined as the delivery of one or more living infants (≥22 weeks gestation or birth weight more than 500 g). Cumulative live birth refers to the sum of deliveries with at least one live birth in the current IVF cycle including all fresh and/or frozen embryo transfers until one live birth was delivered or until all embryos were used, whichever occurs first (18). Biochemical pregnancy was defined as serum beta-hCG≥10 IU/L 14 days after embryo transfer. Biochemical pregnancy loss was defined as a pregnancy diagnosed only by the detection of beta-hCG in serum without any ultrasonographical evidence. Clinical pregnancy was defined when at least one gestational sac was observed on ultrasonography 7 weeks after embryo transfer with or without heart beat, including ectopic pregnancy. Multiple pregnancies were defined as a pregnancy with two or more gestational sacs or positive heart beats 7 weeks after embryo transfer. Miscarriage was defined as the spontaneous loss of an intra-uterine pregnancy prior to 22 weeks gestation. Stillbirth was defined as fetal death occurring during late pregnancy (at 22 completed weeks of gestational age and later) or during childbirth. Preterm birth (PTB) was defined as births that took place before 37 weeks gestation. Low birth weight (LBW) and macrosomia were identified as birthweight <2500g and >4000g, respectively. Small for gestational age (SGA) was identified as birthweight <10th percentile and large for gestational age (LGA) was identified as birthweight >90th percentile, based on Chinese reference singleton newborns stratified by gestational age and sex at birth (19).

The cut-off point of BMI ≥24kg/m² and BMI ≥28kg/m² were adopted for overweight and obesity respectively as recommended for the Chinese population (20). Other variables which may affect the outcomes of IVF/ICSI were also included in the analysis, including female age, educational level, type of infertility, factors of infertility, number of previous IVF cycles, duration of infertility, stimulation protocol, duration of stimulation, total gonadotrophin dose, number of oocytes retrieved, fertilization method, type of embryo transfer, numbers of embryos transferred, and endometrial thickness on embryo transfer day.

Demographic characteristics of the participants were described as means ± standard deviations (SD) for continuous variable, when they met normal distribution, otherwise, median and interquartile ranges were reported. Counts (percentages) were presented for categorical variables. Students’ t-test was applied to compare the difference between two groups for normally distributed variables. Chi-squared tests were used to compare qualitative data, and Fisher’s exact test was applied when the expected frequency was less than five. The dataset included in this study had no missing data. Data management and statistical analysis was performed using SAS statistical analysis software, version 9.4 (SAS Institute, Inc). Two-sided p<0.05 were considered as statistically significant.

The optimal cut point for female WHR was determined by evaluating the area under the receiver operating characteristic (ROC) curve in univariable logistic regression model with a live birth event as a dependent variable and the female WHR (continuous variable) as an independent predictor. The optimal cut point (0.783) was selected, which has the highest Youden Index. Then the female WHR was categorized into low (0.62-<0.783) and high (0.783-<1.09). Univariable logistic regressions and multiple logistic regression models were performed to determine the impacts of female WHR on pregnancy outcomes including live birth, cumulative live birth and miscarriage. The multiple logistic regression models for live birth and miscarriage were adjusted for female age, female education level, female BMI, type of embryo transfer, endometrial thickness, type of infertility, factors of infertility, number of IVF cycle, stimulation protocol, fertilization method, stage of embryo transferred, number of embryo(s) transferred, male WHR. The multiple logistic regression models for cumulative live birth were adjusted for female age, female education level, female BMI, type of infertility, factors of infertility, number of IVF cycle, stimulation protocol, fertilization method, and male WHR. The relationship between female BMI and pregnancy outcomes was analyzed in the same way.

We categorized female WHR as low and high, and weight as normal (BMI: 18.5-<24kg/m²) and overweight (BMI ≥24kg/m²) separately. Then we created a 4-category variable to represent the joint WHR and BMI. In the analysis of joint WHR and BMI with live birth, cumulative live birth and miscarriage, participants with low WHR and normal weight were set as the reference group, and similar covariates were adjusted in the logistic regression models. The area under the curve (AUC) of the female waist-hip ratio with that of BMI was performed by univariable regression models. The interaction between female BMI and waist-hip ratio was analyzed according to the multivariate regression.

In order to confirm the effects of central obesity on IVF/ICSI outcomes, WC was also analyzed in the same way with WHR, while the optimal cut-off value of waist circumference was 74.5cm. Then the female WC was categorized into low (56.0-<74.5cm) and high (74.5-<110.0cm).

This study showed that Chinese women with WHR values above 0.783 had lower live birth rate with a higher miscarriage rate and lower cumulative live birth rate compared with women having WHR less than 0.783, independently of BMI. However, there was no significant associations between BMI and IVF pregnancy outcomes. These results suggest that WHR may be a better predictor of adverse outcomes than BMI in women who underwent IVF/ICSI treatments.

Evidence suggests that obesity has a negative impact on both the follicle environment and embryo development, which may contribute to impaired endometrial decidualization and placental abnormalities (21–23). The levels of free fatty acids, markers of inflammation, and insulin resistance are increased in the follicular environment of women with obesity who underwent IVF (24). In addition, central obesity disrupts the stability of the hypothalamic-pituitary-ovarian axis by affecting the secretion of leptin, reactive oxygen species and other adipokines, which further exacerbates metabolic and reproductive abnormalities (25). Moreover, abdominal obesity may lead to an imbalance in the secretion of sex hormones, especially androgens, which is known to cause infertility in women (26).

As the most widely used indicator of obesity, BMI is closely related to female fertility (5, 27). However, consistent with previous studies (6–8), there were no significant associations between elevated BMI and IVF outcomes in our study. For example, the OR between BMI and cumulative live birth were not statistically significant, whether WHR was adjusted in the multivariate regression analysis model or not. These results may be due to the differences in the study population, BMI values, sample size, selection bias or analysis methods. For instance, only 30% of women in this study were overweight or obese and most were of normal weight.

When comparing women with similar BMI, those with central obesity had higher risk of anovulation (28). A retrospective study of Chinese women found that normal weight obesity (normal BMI but a high percentage of body fat) was associated with lower number of retrieved oocytes, fertilized oocytes, cleaved embryos and good-quality embryos (29). Women undergoing IVF with higher WC had lower live birth rate and higher spontaneous miscarriage rate (30, 31). In contrast, the reduction in WC was associated with an increased odds of pregnancy (32). A prospective cohort study observed delayed conception among black women in the United States with high WHR (≥0.85) after controlling for BMI (33). The pregnancy rate of IVF cycles was lower in women with WHR ≥0.80 than those with gynoid body shape, independent of BMI (34). However, the effects of WHR on major IVF outcomes, such as live birth are not clear. Our study showed that women with relatively high WHR (≥0.783) were associated with lower live birth rate and higher miscarriage rate, regardless of BMI. Since WHR was not associated with biochemical or clinical pregnancy rates, our data suggest that lower live birth rate was due to higher miscarriage rate in women with central obesity. Consistently, higher WC was associated with lower live birth rate after adjusting for BMI and other confounding factors in this study. We further compared the pregnancy outcome AUCs of the female waist-hip ratio with that of BMI. It turned out that the waist-hip ratio (high vs low) showed the highest AUCs. And there was no interaction between WHR and BMI regardless of pregnancy outcomes, which suggested that WHR appears to be a better predictor of female fertility treatment outcomes.

Interestingly, we found that the number of ampoules of gonadotrophin (both initial and total) administered was significantly higher in women with higher WHR, which had not been previously reported. Souter et al. had previously reported heavier women tended to need higher dose of gonadotropins during the stimulation protocol (35). And the mean number of oocytes collected was negatively related to female body weight (6), which is consistent with our findings.

Since even non-obese Asians show increased health risk associated with body fat percentage, common BMI indicators do not apply to Asians. We used the Chinese criteria of overweight (BMI ≥ 24 kg/m²) and obesity (BMI ≥ 28 kg/m²) in this study (20), which is higher than the WHO criteria for Asians. Previous studies have reported different thresholds for WHR threshold ranging from 0.80 to 0.85 in women (36). However, there is no consensus for WHR threshold in different ethnic populations. In our study, we observed a lower WHR threshold (≥0.783) in this cohort, indicating that Chinese women may have a lower WHR threshold as an index of central obesity. When taking both WHR and BMI into account, we found that women of normal weight but high WHR showed statistically significant reduction in fertility, which suggested that WHR rather than BMI was associated with adverse IVF outcomes for Chinese women.

With substantial improvements in cryopreservation and frozen embryo transfer techniques, cumulative live birth rate has been considered a better indicator of the quality and success of IVF by both the patients and the clinicians (37). As the yield of oocyte increases, the cumulative live birth rate increased significantly. Conversely, the reduction of oocyte quality was associated with lower cumulative live birth rate (38). Maternal obesity can lead to a decline in oocyte quality (39). However, the relationship between central obesity and cumulative live birth rate had never been reported before. Our findings indicated that women with higher WHR were associated with lower cumulative live birth rate.

Maternal overnutrition will change the environmental conditions in utero, leading to altered epigenetic modifications, which further affects offspring metabolism (40). Increasing evidence suggests that children of obese mothers have an increased risk of obesity, partly due to maternal overnutrition leading to an oversupply of fatty acids and glucose to the fetus (41, 42). Other studies have also reported that maternal overweight/obesity is associated with an increased risk of high birthweight and subsequent overweight/obesity and fatty liver in offspring (43, 44). For every 0.1 unit increase in maternal WHR, the birthweight of offspring increased by 120 g (45). However, the effects of maternal WHR on IVF offspring have not been reported in the literature. Although we observed some minor differences in the results, it was not statistically significant and this may be due to the low percentage of obesity in this study.

This study has several strengths. To our knowledge, this is the first study mainly focused on the association between WHR and IVF/ICSI outcomes including live birth and cumulative live birth among Chinese women who underwent fresh or frozen embryo transfer. Different from the variable definition of “pregnancy rate”, live birth rate is the preferable standard for evaluating IVF outcomes. Besides, cumulative live birth rate is an important indicator that reflecting the status of all oocytes obtained from a single IVF/ICSI cycle, and the ultimate goal for patients. Notably, two thirds of women in this study were of normal weight, and our findings demonstrated that WHR is an independent factor that is associated with adverse IVF outcomes including the high risk of miscarriage and lower odds of live birth and cumulative live birth. Moreover, the anthropometric and covariate data were collected prior to ovarian stimulation, and most of patients (about 86%) subsequently underwent embryo transfer cycles. Additionally, to adjust for major confounders, multivariable logistic regression analysis was used to minimize bias.

There are some limitations in this study. The main limitation is that the data came from a single IVF center as a retrospective study, which could only provide limited individual covariate information. Secondly, although WHR makes it easier to measure abdominal obesity than as compared to the complexity and high cost of DEXA, it is an anthropometric indicator and is almost constant for small weight fluctuations, and there may be measurement errors. Thirdly, for some patients who underwent frozen embryo transfers, there are variable time intervals between embryo transfer and oocyte retrieval. During this time, women’s WHR might have changed, thus might introduce error for the pregnancy outcomes estimation. Fourth, in clinical practice various protocols were utilized in the IVF/ICSI process, which might contribute to the potential biases in the study.