Association between ambient air pollution and in vitro fertilization pregnancy outcomes in Taiyuan: A retrospective study in a heavy polluted city in northern China

Background Globally, air pollution has a signicant impact on human health. However, the effects of air pollution on pregnancy outcomes in patients undergoing in vitro fertilization (IVF) have not been fully understood. In this study, we analyzed the effects of air pollution on IVF pregnancy outcomes in Taiyuan, which is a heavy polluted city in northern China. Methods 516 patients who underwent rst fresh IVF cycle were enrolled in the retrospective study from January 1, 2015 to May 31, 2020. We collected medical record data from the electronic medical record system and daily average air pollution data from air quality monitoring station. Logistic regression was used to analyze the relationship between six atmospheric pollutants (PM 2.5 , PM 10 , O 3 , NO 2 , SO 2 , CO) and air quality index (AQI) and IVF pregnancy outcomes (biochemical pregnancy and clinical pregnancy) in different exposure periods. Results The results indicated that exposure to NO 2 was negatively associated with the odds of biochemical pregnancy and clinical pregnancy, whereas exposure to O 3 presented positive association. Furthermore, we also found that AQI was negatively associated with IVF pregnancy outcomes. Our ndings suggested that exposure to ambient air pollution during any period may have an impact on IVF pregnancy outcomes, and poor air quality is more likely to reduce clinical pregnancy rates. retrieved were 8.77 mIU/ml, 4.64 mIU/ml, 70.96 pg/ml, and 10.78, respectively. For IVF outcomes, the biochemical pregnancy rate and clinical pregnancy rate were 255 (49.4%) and 217 (42.1%), respectively.


Introduction
With the rapid industrialization, urbanization and motorization, air pollution problems are becoming more and more serious, and the effect of air pollution on health has attracted more attention. Air pollutant is a complex mixture of gaseous components, solid and liquid suspended particles in the air [1], mainly includes particulate matter ≤ 10 µm in aerodynamic diameter (PM 10 ), particulate matter ≤ 2.5 µm in aerodynamic diameter (PM 2.5 ), sulfur dioxide (SO 2 ), carbon monoxide (CO), nitrogen dioxide (NO 2 ) and ozone (O 3 ). Several studies have shown that air pollution has a variety of negative effects on human health, such as cardiovascular diseases [2,3], respiratory diseases [4,5], type 2 diabetes mellitus [6], mental disease [7,8], and so on. In recent years, more evidences showed that there was a correlation between ambient air pollution and human reproductive outcomes, such as infertility, preeclampsia, miscarriage and preterm delivery [9,10].
Infertility is a global problem. It is reported that about 10% of the world's population are infertile [11]. Among couples of childbearing age in China, the incidence of infertility has reached 25%, and it is on the rise [12]. With the development of human assisted reproductive technology (ART), it has become an important mean for treatment of infertility. More than 7 million children have been born worldwide since 1978 using ART [13]. Previous studies had reported that both long-term and short-term exposure to air pollutants were associated with decreased probability of the rate of implantation, biochemical pregnancy, clinical pregnancy, and live birth [14][15][16][17]. However, the present results are inconsistent. Taiyuan is a typical resource-based city with high pollution, which is rich in coal resources. The relationship between air pollution and pregnancy outcomes in other areas could not directly re ect the status in Taiyuan due to its unique air pollution characteristics. Hence, it was of great signi cance and urgency to further investigate the effects of ambient air pollutants on IVF pregnancy outcomes in Taiyuan.
Air quality index (AQI) is an index that is used for reporting daily the air quality of the environment where we live [18]. However, no studies reported the effects of AQI on IVF pregnancy outcomes. To further investigate the impact of six ambient air pollution (PM 10 , PM 2.5 , SO 2 , CO, NO 2 and O 3 ) and AQI on IVF pregnancy outcomes, we conducted a retrospective study. The women who received IVF treatment at Center of Reproductive Medicine of Children's Hospital of Shanxi and Women Health Center of Shanxi in Taiyuan were enrolled into the study from 2015 to 2020. Our results would be helpful to further understand the impact of air pollution on pregnancy outcomes of patients underwent IVF in a heavy polluted city in northern China, and provide some theoretical support for the prevention and treatment of infertility patients.

Study design and subjects
Clinical data from patients who underwent the rst fresh IVF cycle were obtained at the Center of Reproductive Medicine of Children's Hospital of Shanxi and Women Health Center of Shanxi between January 1, 2015 and May 31, 2020. In this study, the participants were collected depending on the patients ' residential location. We assumed that they did not relocate during IVF treatment. Additional selection criteria include: 1) patients underwent their rst fresh cycle, 2) patients used the conventional controlled ovarian hyperstimulation (COH) protocols, and 3) patients' complete medical record data were accessible. Finally, we included 516 IVF patients for analysis.

IVF procedure
In general, IVF procedure is divided into four stages: COH, oocyte retrieval, embryo transfer, and pregnancy test. Firstly, COH protocol was carried out to induce ovulation according to the ovarian response, age and ovarian reserve, in order to promote the development of follicles. Human chorionic gonadotrophin (hCG) was injected when two or more dominant follicles reached 18 mm or more in diameter. Oocyte retrieval was conducted through transvaginal ultrasound-guided follicle aspiration. Oocytes were fertilized by a conventional IVF or ICSI methods. Embryo transfer was performed 3-5 days after oocyte retrieval. Luteal support was started on the day of oocyte retrieval. Biochemical pregnancy is diagnosed by a serum β-hCG greater than 30 mIU/ml at 2 weeks (14 days) after the embryo transfer. Clinical pregnancy is diagnosed by a gestational sac with fetal heart beat during ultrasound examination after ve weeks (35 days) of embryo transfer.

Air pollution data
The daily air pollution data in Taiyuan were obtained from air quality online analysis platform between January 1, 2015 and May 31, 2020 (http://www.aqistudy.cn/historydata). The air pollution data mainly include the concentration of six air pollutants (PM 2.5 in μg/m 3 , PM 10 in μg/m 3 , SO 2 in μg/m 3 , NO 2 in μg/m 3 , CO in mg/m 3 , O 3 in μg/m 3 ) and AQI. Based on these data, we calculated average ambient air pollutant concentration during longterm exposure to air pollutants (period 1, period 2 and period 3) and IVF treatment cycle (period 4, period 5, period 6, and period 7). As shown in Fig. 1, these stages are the following: period 1 (85 days before the oocyte retrieval to ultrasound to determine pregnancy), period 2 (85 days before the oocyte retrieval to serum hCG test), period 3 (85 days before the oocyte retrieval to oocyte retrieval), period 4 (gonadotropin (Gn) start to oocyte retrieval; period 5 (oocyte retrieval to embryo transfer), period 6 (embryo transfer to serum hCG test), and period 7 (embryo transfer to ultrasound to determine pregnancy).

Statistical analysis
Descriptive analyses were presented as count (%) for categorical variables or as mean ± standard deviation (SD) for continuous variables, respectively. Air pollutants were summarized by the median and quartiles (25th-50th-75th quartile). Spearman correlation analysis was performed to examine the correlation among the pollutants. Univariate and multivariate logistic regression analyses was used to evaluate the associations between ambient air pollution exposures and IVF pregnancy outcomes. Odd ratios (ORs) and 95% con dence intervals (Cls) were used to report the result. For multiple logistics analysis, we divided the exposure levels of air pollutants in different expose periods into four groups (Q1: low levels, Q2: medium levels, Q3: high-medium levels, Q4: high levels) according to the quartile, and Q1 group was used as a control, adjusting for age, BMI, duration of infertility, type of infertility, causes of infertility, IVF/ICSI, basal follicle-stimulating hormone (FSH), basal luteinizing hormone (LH), basal estradiol (E2), number of oocytes retrieved for analysis. All statistical analyses were performed using SPSS 22. A p-value of < 0.05 was considered statistically signi cant.

Study population characteristics
This retrospective population-based cohort study consisted of 516 patients with IVF from January 1, 2015 to May 31, 2020 in Taiyuan. The descriptive statistics of study population was given in Table 1. Among the 516 women, 297 (57.6%) occurred primary infertility and 219 (42.4%) occurred secondary infertility. In addition, 382 (74.0%) underwent IVF and 134 (26.0%) underwent ICSI. Their mean age, mean BMI and mean duration of infertility were 31.48 years, 22.83 kg/m 2 and 3.5 years, respectively. The main causes of infertility were tubal and uterine factor 270 (52.3%). Their mean basal hormone levels (FSH, LH and E2) and mean number of oocytes retrieved were 8.77 mIU/ml, 4.64 mIU/ml, 70.96 pg/ml, and 10.78, respectively. For IVF outcomes, the biochemical pregnancy rate and clinical pregnancy rate were 255 (49.4%) and 217 (42.1%), respectively.

Air pollution exposure characteristics
The average ambient air pollutant concentration for six air pollutants and AQI during different exposure periods were exhibited in Table 2. In addition, we used the Spearman's correlation to examine the correlations among the pollutants. The results showed that concentrations of the air pollutants examined (PM 10 , PM 2.5 , SO 2 , CO, NO 2 ) were positively correlated (Rang: 0.23 to 0.89), and the correlations had highly signi cant in statistics (P < 0.01). However, the concentration of O 3 was negatively correlated (Rang: −0.13 to −0.80) with signi cant difference (P < 0.01).
Association between air pollution and pregnancy outcomes As shown in Table 3, we performed univariate logistic regression analysis of the relationship between 6 air pollutants and AQI and IVF pregnancy outcomes (see Additional le 1 and 2). We found that 6 air pollutants and AQI were not associated with the IVF pregnancy outcomes (P 0.05). Similarly, we explored the estimated effects by air pollutants quartiles during different expose periods with Q1 as the reference, adjusting for age, BMI, duration of infertility, type of infertility, causes of infertility, IVF/ICSI, basal FSH, basal LH, basal E2, number of oocytes retrieved. It indicated that there was a certain relationship between air pollutants to exposure and IVF pregnancy outcomes.
The result of biochemical pregnancy was shown in Table 3. During period 4, medium levels (Q2) and high levels (Q4) of NO 2 were signi cantly associated with reduced odds of biochemical pregnancy. High levels (Q4) of NO 2 exposure during period 7 was signi cantly associated with reduced odds of biochemical pregnancy and decreased in a dose-dependent manner as exposure levels quartiles increasing (P for trend = 0.031). On the contrary, medium levels O 3 (Q2) during period 4 and 5 and high-medium levels O 3 (Q3) during period 5 were associated with increased odds of biochemical pregnancy.
The results for the clinical pregnancy rate was shown in Table 4. High-medium levels NO 2 (Q3) during period 1, 3 and 6 and medium levels (Q2) during period 4 were signi cantly associated with reduced odds of clinical pregnancy. In contrast, high-medium levels O 3 (Q3) during period 1 and 5 were associated with increased odds of clinical pregnancy. For AQI, high-medium levels (Q3) during period 1 and high levels (Q4) during period 4 were signi cantly associated with reduced odds of clinical pregnancy. The odds of clinical pregnancy showed a decreasing trend (P for trend = 0.005) with AQI quartiles increasing during period 4.

Discussion
With the deterioration of air quality in recent years, more and more people pay attention to the health effects of air pollution. In this retrospective study, we explored the association between 6 air pollutants and AQI and IVF pregnancy outcomes during long-term exposure (period 1, period 2 and period 3) and IVF treatment cycle (period 4, period 5, period 6, and period 7) in Taiyuan. The results suggested that lower pregnancy outcomes in IVF cycles might be linked with worse air environment. The results further supported previous studies that air pollution exposure had the negative in uences on IVF pregnancy outcomes.
Previous studies have explored the relationship between air pollution and IVF pregnancy outcomes. A study from three IVF laboratories in the Northeastern United States found that the increase of NO 2 concentration was signi cantly associated with a low chance of pregnancy and live birth [17]. And another study, Boulet et al. reported that exposure to O 3 was weakly positively associated with implantation rate and live birth rate, while exposure to PM 2.5 was not associated with any of IVF treatment outcomes [16]. Perin et al. indicated that high levels of ambient PM during the follicular phase had a higher risk of pregnancy loss in Brazilian women undergoing IVF [19]. Choe et al showed that increase of PM 10 , NO 2 and CO were associated with a decreased probability of intrauterine pregnancy in Korea women after adjusting for individual risk factors [20]. Also from South Korea, Kang et al. showed that exposure to NO 2 , PM 10 , and PM 2.5 were associated with decreased pregnancy rates in women undergoing fresh IVF-ET treatment from a large, long-term database [15].
Additionally, there have been related reports in China. In Xiamen, Wang et al. retrospectively analyzed the effect of outdoor air pollution on IVF outcomes of frozen-thawed embryo transfer (FET) and fresh embryo transfer in the same period. They found that exposure to high levels of SO 2 and O 3 were signi cantly associated with low live birth rates in frozen embryo transfers but had no effect on fresh embryo transfer [21]. In Shenyang, Qiu et al. observed that an inter quartile range (IQR) increasing of O 3 during Gn start to oocyte retrieval and one day before oocyte retrieval was associated with a lower intrauterine pregnancy rate, while there was an inverse association of intrauterine pregnancy with CO two days prior to oocyte retrieval [22]. In Chengdu, Zeng et al. found that expose of PM 2.5 , PM 10 , NO 2 , SO 2 and CO were negatively associated with the odds of biochemical pregnancy and clinical pregnancy except O 3 , particularly < 35 years younger women [14]. The inconsistency of the above conclusions could be explained as follows: 1) the composition, levels and sources of air pollution vary due to different countries and regions; 2) the individual susceptibility is different due to different genetic backgrounds and lifestyles; 3) most of the above studies are retrospective cohort studies, which may increase the possibility of selection bias, failure to control for selection bias potential confounding factors and sample size.
Taiyuan, is a typical heavy industry city characterized by coal-burning pollution, mainly comes from the exhaust emissions of the heavy industries and coalred heating in winter [5]. With the rapid development of urbanization, motor vehicle exhaust emissions have become one of the main sources of air pollution in Taiyuan city [23]. Additionally, Taiyuan is surrounded on the three sides by mountains that inhibit the dispersion of pollutants. The special basin terrain character and temperature inversion in winter make it di cult to disperse air pollutants. Therefore, the air pollution characteristics of Taiyuan was formed that the dominated pollutant by particulate matter and SO 2 , accompanied by a large amount of NO 2 and O 3 . Similar to the above results, we also found that air pollutants were associated with reduced biochemical pregnancy rates and clinical pregnancy rates among IVF patients in Taiyuan. NO 2 is a tra c-related air pollutants that was mainly released from vehicle emission and coal combustion [24]. In this study, we found that there was a signi cant negative correlation between exposure to NO 2 and biochemical and clinical pregnancy, while exposure to O 3 had the inverse relationship. The results of this study are similar to Zeng's [14]. In addition, Slama et al. reported that high concentrations of PM 2.5 and NO 2 were associated with lower pregnancy rates in women [25]. In Balb/c mice, Mohallem et al. also found that the failure rate of implantation increased and the live birth rate decreased exposed to high levels of NO 2 and PM 2.5 [26]. In addition, O 3 is a secondary pollutant that is not emitted directly into the atmosphere, but is produced by photochemical reactions between NO 2 and hydrocarbons [27]. Therefore, the rise of O 3 re ects good air quality to some extent. This not only explains that O 3 is positively correlated with pregnancy outcome, but also explains that O 3 is negatively correlated with other pollutants and AQI.
AQI as a quantitative index is to describe air quality status. The higher the AQI value, the more serious the air pollution and it has greater risk to human health [28]. At present, there were no data to analyze the relationship between AQI and IVF pregnancy outcomes. In the study, we found that there was a dose-dependent relationship between AQI and clinical pregnancy rate, and the clinical pregnancy rate decreased with AQI quartiles increase. The results showed that the worse the air quality, the lower the clinical pregnancy rate.
In this study, AQI was rstly used to explore the relationship between air quality and IVF pregnancy outcomes. As a heavy polluted city in northern China, we further explored the effect of air pollutants during different exposure periods on IVF pregnancy outcomes in Taiyuan.
However, certain limitations of this study should be mentioned. First, the level of individual air pollution exposure in this study do not accurately represent individual exposure level. In this study, the data of averaging daily pollutant concentrations was collected from an air pollution monitoring station. In addition, it is di cult to directly measure individual air pollution exposure due to its expensive cost and the patients' compliance. Furthermore, we only conducted the single-pollutant models based on six air pollutants, without the additive and/or interactive effects of two or more pollutants. As an environmental evaluation indicator based on six pollutants, we explore the relationship between AQI and IVF pregnancy outcomes. Third, the number of participants and potential confounders were limited due to retrospective study. Therefore, a prospective study with a large sample size will be more conducive to draw more reliable conclusions.

Conclusions
In summary, our ndings suggested that air pollution has a certain impact on IVF pregnancy outcomes, especially NO 2 and O 3 . In addition, the results of AQI further indicated that poor air quality was more likely to reduce clinical pregnancy. Thus, to strengthen the protection of IVF infertility patients from air pollutants is bene cial to increase the biochemical pregnancy rate and clinical pregnancy rate.