2. Patients and Methods
Study design and participants. We recruited 134 healthy pregnant
women at the “Clinica Mangiagalli”, Fondazione IRCCS Ca’ Granda
Ospedale Maggiore Policlinico, Milan, Italy. The women were randomly
selected from individuals who were attending prenatal healthcare clinics
during the 11–12th week of pregnancy. Exclusion criteria included a
history of illicit drug use, diabetes, hypertension, previous pregnancy
with pre-eclampsia/eclampsia or gestational hypertension, and current
use of acetylsalicylic acid or low-molecular-weight heparin. Information
about demographics and lifestyle characteristics of the mother, such as
smoking habits and alcohol consumption, were collected. An informed
consent form was signed by all participants and the study was approved
by the ethics committee of the Fondazione IRCCS Ca’ Granda Ospedale
Maggiore Policlinico (approval number 681/2017).
Clinical and laboratory measurements. Body weight and height were
determined on a standard scale. Body mass index (BMI) was expressed as
Kg/m2. Systolic and diastolic blood pressure (SBP and
DBP, respectively) were taken on the left arm using a mercury
sphygmomanometer (mean of two measurements taken after 5 min of rest).
Plasma lipids/lipoproteins and glucose were measured by certified
enzymatic techniques on a Roche c311 autoanalyzer. Lipoprotein (a)
[Lp(a)] levels were measured by immunoturbidimetry on a Roche c311
autoanalyzer. Standard evaluations for early pregnancy in Italy are
serum pregnancy-associated plasma protein-A (PAPP-A), α-fetoprotein, and
human chorionic gonadotropin (hCG). These parameters were measured at
11–12 weeks of gestation. Gestational age was calculated from the last
menstrual period, and was verified by ultrasound parameters. In
particular, fetal crown-rump length was used to estimate gestational
age, and women were included if this parameter ranged between 45 and 84
mm.
Enzyme-linked immunosorbent assay (ELISA). Plasma PCSK9
concentrations were measured by a commercial ELISA kit (R&D Systems,
MN). All patients fasted overnight and had blood sampled at around
09:00, thus minimizing any possible confounding effects of circadian
variation in PCSK9 levels. In brief, samples were diluted 1:20 and
incubated onto a microplate pre-coated with a monoclonal human
PCSK9-specific antibody. Sample concentrations were obtained by a
four-parameter logistic curve-fit, with a minimum detectable PCSK9
concentration of 0.219 ng/mL (25). Intra-
and inter-assay CVs were 3.8% and 6.2%, respectively.
Air pollutant assessments. Daily air pollutant
(PM10, PM2.5, and NO2,)
concentrations were derived from the archives of the Regional
Environmental Protection Agency (ARPA Lombardy). This organization
collects data at a regional scale using the FARM (Flexible Air quality
Regional Model) chemical-physical model of air quality
(26). This model is a three-dimensional
Eulerian model that simulates the dispersion and chemical reactions of
atmospheric pollutants. The estimated levels of daily
PM10, PM2.5, and NO2concentrations were assigned to each subject for the day of evaluation
and 14 days before blood was sampled. We also calculated the average
exposure from the first week before the clinical visit and 12 weeks
earlier (i.e., weeks 0–1 being the mean over the first week of
exposure and weeks 0–12 being the mean over the 12 weeks before the
visit. All participants were assigned pollutant levels that were
estimated in the Municipality of Milano, as 93% of the women lived or
worked there.
Statistical analysis. Descriptive statistics were performed on
all variables. Continuous variables were expressed as the mean ±
standard deviation (SD) or as the median with first-, and third-quartile
(Q1–Q3), as appropriate. Categorical data were reported as frequencies
with percentages. Descriptions of each exposure variable were given by
the means of box-plots, describing pollutants at each averaged time
window. We applied univariate and multivariable linear regression models
to evaluate the relationship between pollutant exposure (for each
averaged one-week period from week 0–1 to week 0–12) and circulating
PCSK9 levels. Each model was tested for normality and linearity. All
potential confounders were included in the multivariate model after
verifying the presence of an association in a univariate model. Best
model selection was based on the minimization of the Akaike information
criterion and maximization of the explained variance of the model. The
final models were adjusted for low-density lipoprotein cholesterol
(LDL-C), interleukine (IL)-6, fibrinogen, season, BMI, and smoking
habit. Estimated effects are reported as β and standard error (SE)
associated with an increase of 1 unit in each pollutant.
We examined the association between PCSK9 and the variables measured on
the newborn (gestational age at birth, weight, length, cranial
circumference, APGAR score), after adjusting each model for the
pollutants most associated with PCSK9 levels in the multivariate
analysis. Each model was also adjusted for birth mode (urgent caesarean,
elective caesarean, and spontaneous delivery) and for the interaction
between pollutant and PCSK9 concentrations. Using a univariate logistic
regression, we evaluated the odds ratio of urgent cesarean delivery
associated with a 100 mg/dL increase in PCSK9.
We calculated the q-FDR values using the multiple comparison method
based on Benjamini-Hochberg False Discovery Rate (FDR), which takes the
high number of comparisons into account, with a threshold of 0.10 to
detect significance.
A sensitivity analysis was performed using the residential address for
pollutant imputation, with no relevant changes being made to the results
(data not shown). Statistical analyses were performed with SAS software,
version 9.4.