3. Results
Study population . The study population included 134 pregnant women (age 33 ± 4 years). BMI was 22.6 ± 4.2 kg/m2. Pregnancy associated endocrine factors, such as placental growth factor (PLGF) and pregnancy associated plasma protein A (PAPP-A), were in the normal range, as was hCG (Table 1). Signs of diabetes and dyslipidemias were not found. Specifically, median glucose was 86.6 ± 14.8 mg/dL. LDL-C, non-high-density lipoprotein cholesterol (non-HDL-C), HDL-C, and triglycerides (TG) were also within the normal range (97.4 ± 22.1 mg/dL, 117.2 ± 25 mg/dL, 65.5 mg/dL and 99 ± 38.3 mg/dL, respectively). Levels of Lp(a) were 13.7 ± 19 mg/dL (Table 1). Circulating levels of plasma PCSK9 were normally distributed, with a mean of 193.7 ± 54.2 ng/mL (Figure 1). These levels were lower than those previously described by us for non-pregnant women recruited in the same geographical area (27). None of these healthy women were on any drug treatment. The estimated level of exposure to PM10, PM2.5, and NO2, from the first week before the visit (week 0–1) and 12 weeks previously (week 0–12), are depicted in Figure 2. The similarities in pollutant concentrations across the weeks of exposure were clearly observed. Mean PM10 and NO2concentrations remained beneath the annual regional air-quality standards of 40 μg/m3. Mean PM2.5concentrations were slightly higher than annual limits, which are set at 25 micrograms per μg/m3.
Mean fetal crown-rump length at the time of exposure assessment was 62.4±5.2 mm. Nuchal translucency thickness, which is an ultrasound marker for chromosomal and structural abnormalities, was within the normal range in all cases, as well as fetal heart rate and ductus venosus blood flow (Table 1). All pregnancies ended with the live birth of a phenotypically normal neonate, at a mean gestational age of 38.7  1.4 weeks. Neonatal biometric parameters are presented in Table 1.
Univariate data analysis. None of the three air pollutants was associated to circulating PCSK9 levels in the univariate analysis (Supplemental Table 1). As expected, the main lipid parameters related to CV risk, LDL-C (β = 0.605, SE = 0.235, p = 0.011) and non-HDL-C (β = 0.425, SE = 0.210, p = 0.045) were positively associated with PCSK9. Inflammatory markers, which are linked to the initiation and progression of atherosclerosis, were also evaluated. PCSK9 levels were positively associated with the IL-6 (β = 3.447, SE = 1.711, p = 0.046) and fibrinogen (β = 0.170, SE = 0.083, p = 0.043) levels. No relationship was obtained with the high-sensitivity C-reactive protein and with the adhesion molecules, ICAM and VCAM. Newborn features, e.g.,crown-rump length, fetal heart rate, and cranial circumference were not associated with PCSK9 levels (Supplemental Table 1).
Multivariable data analyses. To define the exposure window to air pollutants that was most effective in modifying PCSK9, we investigated how different time lags were associated with PCSK9 levels.
We observed a positive significant effect of PM10exposure for all time lags. The effect on PCSK9 levels was maximal, as we considered the mean of the entire gestational period, i.e.,0–12 weeks (Table 2). In particular, for each 1-µg/m3increase in PM10 concentration, we observed a significant increase in PCSK9 levels (β = 1.903, SE = 0.733, p = 0.011). This effect was also confirmed for NO2 exposure, as every unit increase in NO2 led to a 2.265 ng/mL rise in PCSK9 levels (β = 2.265, SE = 1.002, p = 0.026). For PM2.5 exposure, a significant positive association was only detected with the 0–6 and 0–10-week time lags.
Association of PCSK9 concentrations with gestational age for different levels of exposure to pollutants. No association was found between PCSK9 measured at the first trimester of pregnancy and features of newborns at birth, such as weight and length at birth, cranial circumference at birth, and APGAR score (Supplemental table 1). Nevertheless, when the interaction between PCSK9 concentrations and gestational age at birth was taken into account, we observed a strong modifying effect of air pollutants, particularly PM2.5. For different PM2.5 levels (15 mg/m3, 24 mg/m3, 42 mg/m3, and 55 mg/m3, respectively; 25°, 50°, 75°, and 95° percentile), the association was significant at the highest PM2.5 concentrations (i.e., those in the 75th and 95th percentiles; Figure 3). For example, at a PM2.5 concentrations of 42 mg/m3, we observed an advance in delivery date of approximately 1 week every 100 ng/mL rise in circulating levels of PCSK9 (β = -0.810, SE = 0.332, p = 0.0164). The steepness of the association was more evident when the highest quartile of PM2.5 was considered (i.e., delivery advanced by 1.28 weeks for every 100 ng/mL change in PCSK9 levels; β = -1.282, SE = 0.498, p = 0.012). Overall, at fixed PM2.5 concentrations of 42 mg/m3 and 55 mg/m3, for every 100 ng/mL increment in PCSK9, the gestational age decreased by 2.3% to 3.7% weeks. This finding was also supported by the observation that the odds ratio (OR) of urgent cesarean delivery associated with a 100 mg/dL rise in PCSK9 was 2.99 (95% CI 1.22–6.57). A similar trend was found when PM10 and NO2 were considered (Supplemental table 3).