The APPO study is a prospective hospital-based cohort study designed to investigate the effects of PM10 and PM2.5 exposure on mothers and fetuses. Pregnant women were included in this study. To develop the specific hypotheses relevant to PM exposure and the maternal and fetal effects, our team reviewed a list of hypotheses from previous studies, including published articles, reports, and textbooks. Based on this review, key research hypotheses were established, and medical, social, and other factors that can affect pregnant women and fetuses were considered. Thus, we generated key hypotheses regarding pregnancy, childbirth, and neonatal prognosis associated with environmental exposure (Table 1).

The APPO study plans to investigate the effects of particulate matter on mothers and fetuses by recruiting more than 1,200 participants from January 2021 to December 2023. Six university hospitals (Kangwon National University Hospital, Keimyung University Dongsan Medical Center, Korea University Guro Hospital, Severance Hospital, Ewha Womans University Mokdong Hospital, and Ewha Womans University Seoul Hospital) participated in 2021, and Ulsan University Hospital was included in 2022. The hospitals were located in a metropolitan area (Seoul), an industrial complex (Guro-gu in Seoul, Ulsan), and a mountainous area (Gangwon-do) to fully reflect the characteristics of different regions. Seoul was selected as an important area for this study because of its large population, high traffic volume, severe air pollution, and several apartments [24]. Moreover, Ulsan was selected as another research area as the largest industrial city in South Korea [24].

To investigate the effects of particulate matter on pregnancy complications and adverse pregnancy outcomes, early pregnant women without underlying diseases were selected for the study. The inclusion and exclusion criteria are presented in Table 2. Participants were excluded if they withdrew their consent or could not provide information on delivery because of follow-up loss. When the participants visited the hospital for regular evaluation in the 1st, 2nd, and 3rd trimester of pregnancy, we collected information related to pregnancy complications through questionnaires, ultrasound sonography, and blood and urine tests. After delivery, umbilical cord blood and placental tissue were collected, and information regarding pregnancy outcomes was obtained from medical records. A flowchart of the APPO study is shown in Fig. 1.

Biological samples collected include 5 mL of maternal venous blood and 15 mL of urine in each trimester of pregnancy; in addition, 5 mL of umbilical cord blood and 2×2×2 cm placental tissue were collected after delivery. Placental tissue was mainly collected at the Ewha Womans University Mokdong Hospital, a representative institute. Biological samples were collected when participants visited for regular evaluation to avoid discomfort, and all biological samples were collected after obtaining informed consent from the participants. Whole blood samples collected in ethylene-diamine-tetraacetic acid tubes were dispensed into cryo-tubes (1.0 mL), and urine was stored in a cryo-tube (10 mL). Samples for long-term storage were transferred to the institution (Seegene Medical Foundation, Seoul, Korea) on the same day by refrigeration to prevent deterioration. The storage samples were labeled with a unique number, collection date, and sample name; afterward, they were systematically stored and managed in a laboratory freezer (−80°C) after being delivered by the Department of Obstetrics and Gynecology at Ewha Womans University Mokdong Hospital. Specific laboratory test results for each biological specimen are presented in Table 3. Tests for systemic diseases, such as inflammation, thyroid disease, and metabolic syndrome, were mainly analyzed from blood samples, and additional tests, such as oxidative stress, were analyzed from urine samples. 8-hydroxy-2′-deoxyguanosine (8-OHdG) is an oxidized nucleoside of DNA frequently detected in damaged nuclear and mitochondrial DNA; and 8-OHdG is excreted in the urine following DNA repair [25]. There is much evidence that malondialdehyde (MDA) is not only a biomarker of generalized cellular oxidative stress, but may also be a risk factor for specific diseases [25]. MDA is the most studied product of polyunsaturated fatty acid peroxidation and is related to lipid oxidation among oxidative stress markers [26]. Moreover, the placenta and umbilical cord blood are mainly used to test the effects in neonates. Black carbon is a major component of the fine dust produced during the combustion of fossil fuels and biomass. It is small enough to be inhaled and deposited directly in the lungs [27]. Therefore, we hypothesized that it can eventually be deposited in the placenta through a blood vessel and observable in the placental tissue. Furthermore, DNA methylation is an epigenetic mechanism related to the fetal programming theory, which states that the environment in utero affects fetal gene expression. Therefore, this study plans to analyze the DNA methylation status in umblical cord blood to predict its effect on the fetus (Table 3).

We collected the following clinical information from each mother’s medical record: prenatal clinical information (cervical length, blood pressure), labor and delivery (delivery date, gestational age, mode of delivery, weight of placenta), pregnancy complications (preterm premature rupture of membrane, dystocia, cephalopelvic disproportion, breech or other abnormal presentation, placenta previa, placental abruption, obstetrical hemorrhage, precipitating labor, PTB, umbilical cord prolapse, eclampsia), neonatal data (sex, height, weight, 1′ and 5′ Apgar score), neonatal complications (meconium aspiration syndrome, fetal asphyxia), and other information (fever, anesthetic use). In addition, oxidative stress index, smoking exposure, carbon particle analysis, and analysis of cord blood DNA methylation are required for risk analysis of particulate matter; therefore, they would be analyzed in high PM10 and PM2.5 concentrations.

We used clinical research and trial management system (iCReaT ver. 2, National Institute of Health, Cheongju, Korea) a web-based system, to input, store, and integrate the aforementioned data. iCReaT was developed to provide evidence-based health care services in an efficient and integrated manner by the National Institute of Health in Korea; it oversees clinical research, including clinical trials, research plan review, project management, data input and extraction, and report provision [28]. Data quality was monitored periodically, the program automatically screened, and additional investigations conducted to obtain information about the questionnaire items left blank or insufficient answers were supplemented. The iCReaT system was suitably configured to handle the special requirements of the APPO study, and a training program for its use was provided for all administrators and researchers. Access to all data was limited to the certified researchers in the study protocol.

In this study, we investigated the current status of PM that pregnant women in Korea are exposed to indoors through a web-based questionnaire (kopen.or.kr), and we collected information on the participants’ sociodemographic characteristics, lifestyle, living environment, and risk perception level of fine dust. This questionnaire was developed by a cohort of housewives, and it was possible to indirectly evaluate exposure to air pollution by examining house occupancy, cooking, and cleaning times, which are factors that generate indoor fine dust, and indoor/outdoor ventilation (number of windows opened). These data were input and stored in the iCReaT system of each institution, and a representative research director integrated and managed the data.

In addition, it was recommended that all participants fill out a time-activity log on the web-based questionnaire. They recorded the information using a time-activity diary, and activity information was collected at 1-hour intervals. The activities were organized according to the following classifications: main activity, extra activity, transportation, indoors, and outdoors. Participants’ time activities were analyzed to identify the indoor and outdoor residence time patterns. The identified pattern was used to calculate individual exposure to indoor and outdoor PM10 and PM2.5.

The outdoor PM concentrations (PM10 and PM2.5) were collected from a nearby urban atmospheric measurement network based on the addresses of the pregnant women who participated in the study. The urban air monitoring station data used in this study were obtained from air Korea (https://www.airkorea.or.kr/web), which is operated by the Korean Ministry of Environment. Air Korea has been disclosing its national real-time air pollution level information via a website since December 2005 [29]. As at 2021, there were 614 stations in 162 cities and counties in Korea for air quality monitoring of sulfur dioxide, nitrogen dioxide, ozone, carbon monoxide, PM10, and PM2.5 [29].

The indoor concentrations of PM10 and PM2.5 were measured by placing a fine dust meter in the living room of each participant’s home. AirGuardK (Kweather Co, Seoul, Korea) was used to measure fine particles (PM10, PM2.5), carbon dioxide (CO₂), volatile organic compounds (VOCs), temperature, humidity, and noise. These measurements were taken online at 1-minute intervals. The measured indoor air quality data are stored in the indoor air quality monitoring platform through the long-term evolution communication network to prevent loss. In addition, the data are collected and stored every minute. Measurements were carried out for a week during each trimester of pregnancy, and the measured concentration value was determined in real time using the internet of things and information and communication technology.

We used a time-weighted average model to estimate the individual PM exposure in pregnant women. The time-weighted average model is a concept in which an individual’s exposure over a 24-hour period can be expressed as the sum of exposure to each microenvironment weighted to a time-activity pattern. Although it was not possible to measure the concentration of air pollutants in all local environments, individual exposure to PM10 and PM2.5, using a time log and time-weighted model, can be predicted [25–28]. The time-weighted model can be expressed by the following equations: C_per=∑(T_indoor×C_indoor+T_outdoor×C_outdoor) where, C_per: predicted personal exposure, T_indoor: indoor activity time rate (indoor activity time/24hr), C_indoor: indoor PM2.5 concentration, T_outdoor: outdoor activity time rate (outdoor activity time/24hr). C_outdoor: outdoor PM2.5 concentration.

In this study, by applying PM10 and PM2.5 concentration values and temporal activity patterns from the time-weighted average model, predicted individual exposure values for pregnant women was obtained. For pregnant women who worked in the office, PM10 and PM2.5 concentrations at the workplace were excluded because of the lack of measurement data. As a result, a time-weighted average model of indoor and outdoor PM10 and PM2.5 concentration values and temporal activity patterns was applied to represent the predicted values for individual exposure.