Using principal component analysis to examine associations of early pregnancy inflammatory biomarker profiles and adverse birth outcomes.

ABSTRACT

OBJECTIVE: Inflammation as a risk factor for preterm birth is well-established. The primary objective of this analysis was to examine whether individual cytokines versus a composite indicator of mid-pregnancy inflammation are significantly associated with risk for adverse birth outcomes. STUDY DESIGN: A multi-site prospective study was conducted in a socio-demographically diverse cohort of 610 pregnant participants. At a study visit between 12 and 20 6/7 weeks' gestation, low-grade inflammation was measured via log-transformed serum concentrations of the biomarkers IFN-γ, IL-10, IL-13, IL-6, IL-8, TNF-α, and CRP. Principal component analysis (PCA) was used to identify underlying dimensions of inflammatory activity from the seven biomarkers measured. Gestational age and birth weight at delivery were obtained from medical chart review. The associations between inflammatory profiles and birth outcomes were assessed via linear and logistic regression models. Results were compared with those from individual inflammatory biomarkers, and model fit was assessed using Akaike's Information Criterion (AIC). RESULTS: Principal component analysis analysis yielded a two-factor solution, with the first factor (IF1) composed of IL-8, IL-10, IL-13, IFN-É£, and TNF-α, and the second factor (IF2) containing IL-6 and CRP. When adjusted for race, education, BMI, smoking status, gestational age at time of blood draw, and study site, a one standard deviation (SD) increase in IF1 remained significantly associated with a decrease in standardized gestational age (ß = -.13, 95% CI -.21, -.05) and an increase in odds of preterm delivery (OR = 1.46, 95% CI 1.13, 1.88) (Table 3). A one SD increase in IF2 was similarly associated with a decrease in standardized gestational age at delivery (ß = -.13, 95% CI -.23, -.04) and an increase in odds of preterm delivery (OR 1.46, 95% CI 1.04, 2.05). Neither IF1 nor IF2 was associated with measures of fetal growth. AIC identified that IL-6 was a slightly better fit for length of gestation compared to either composite measure, though all performed similarly. CONCLUSION: Independent of known sociodemographic risk factors, an elevated mid-pregnancy inflammatory profile was associated with a nearly 50% increase in odds of preterm delivery. The composite performed similarly to IL-6. These results suggest that maternal low-grade inflammation is a risk factor for preterm delivery, and that mid-pregnancy inflammatory biomarkers may be useful in predicting risk for preterm delivery.