2 Methods
2. Study Setting and Population
Our study was a secondary analysis of the Nulliparous Pregnancy Outcomes
Study: Monitoring Mothers-to-Be (nuMoM2b), a large prospective
observational cohort study designed to assess contributors to adverse
pregnancy outcomes. Detailed nuMoM2b protocols were previously published
and are briefly summarized here.(13) Participants in the parent study
were recruited at 8 geographically diverse U.S. sites from 2010-2013 and
were included if they had a singleton pregnancy between 6 weeks 0 days
and 13 weeks days’ gestation and no prior pregnancies lasting 20 weeks
or more. Potential participants were excluded for age <13 years,
3 or more prior miscarriages, suspected fetal malformation at the time
of enrollment, known fetal aneuploidy, conception using a donor oocyte,
multifetal reduction, plan for pregnancy termination, or participation
in an intervention study to influence pregnancy outcomes. Participants
had 4 study visits: during approximately the first, second, and early
third trimesters of pregnancy, as well as one after delivery. For this
secondary analysis, we included all participants with a live birth at
≥24 weeks with available maternal single nucleotide polymorphism (SNP)
array data, derived from unselected maternal blood collection that was
part of the protocol for all parent study participants. Participants
were excluded if they did not complete any of the 3 research ultrasounds
or were missing key variables including fetal sex and BW.
2.2 Outcomes
The primary outcome of this study was association of one’s genetic risk
score for infant BW with race. Race was divided into self-identified
race and genetic ancestry. The secondary outcome of this study was to
assess the relationship and overlap between self-identified race and
genetic ancestry.
Race designations were self-identified from among the following: White,
Black/African American, Asian, Native Hawaiian/Other Pacific Islander,
American Indian/Alaskan Native, Multiracial, and Unknown/not reported.
Genetic ancestry was ascertained using Peddy, a software package
that uses an individual’s DNA to predict the predominant continental
ancestry(14) with the following categorical outputs: AFR, African; AMR,
American (Indigenous); EAS, East Asian; EUR, European; SAS, South Asian;
UNK, unknown. We assessed the distribution of predicted genetic ancestry
within self-identified racial groups.
Maternal DNA was isolated from blood collected at visit 1. Genotyping
was performed using a commercially available kit (Infinium Multi-Ethnic
Global D2 Bead Chip; Illumina), from which SNP arrays were conducted
based on the Genome Reference Consortium human build 38 (CRCh38).(15) 86
BW-associated SNPs that were identified using GRCh37(7) were mapped to
GRCH38 for compatibility, yielding 73 SNPs. Maternal SNP arrays were
used to compute the GRSBW for each maternal participant
using the weighted sum of BW-associated variants present in each person,
such that the score represents the cumulative effect size without
traditional units, expressed as GRS =
(V1*β1) +
(V2*β1) +
…(V73*β73) , whereV1 is variant 1 and β1 is
the effect size for variant 1.
2.3 Statistical Analysis
A log-linear model was used to test the association between maternal
GRSBW with infant BW, controlling for fetal sex and
gestational age at birth. To assess generalizability of the
GRSBW across self-identified racial groups, the
association between GRSBW and BW was assessed for each
self-identified racial subgroup using stratified log-linear models. This
same approach was repeated across groups defined by genetic continental
ancestry. Finally, self-identified race and genetic ancestry were
included as predictor variables in separate log-linear models to test
whether they remained independently associated with BW after controlling
for GRSBW, infant sex, and gestational age.