636: Serum inflammatory analytes as predictors of small for gestational age (SGA) neonates: An expanded analysis

Abstract

Previously we reported a significant association between serum IL-1β (higher), CRP levels (lower) and SGA infants in all trimesters. We sought to determine if these results were useful when grouped together as a screening modality. This was a secondary analysis of a longitudinal multicenter study of singleton gestations. Serum cytokines (IL-1α, IL-1β, IL-6, IL-8, IL-10, TNF-α, CRP) were measured were measured in each trimester. SGA (< 10th percentile) was determined using the Fenton Growth Chart. Group based trajectory (GBT) modeling was used to identify distinct trajectories of cytokine profiles, for those that were significantly associated with SGA status in bivariate analysis. Following the GBT modeling, women were classified as high/low trajectory for each cytokine based on their maximum estimated probability of group membership, and this was used in the logistic model. Cytokine values were log-transformed. Logistic regression was used to evaluate predictors of SGA status, with cytokine trajectories, adjusting for age, first trimester smoking status, and with and without BMI. 317 women underwent screening in the first trimester with delivery of 23 SGA neonates (7%). Serum cytokine data was available for 240, 227 and 214 subjects in the first, second, and third trimesters, respectively. For both IL-1β and CRP, two distinct trajectories were identified. In the logistic model adjusting for age and smoking status, the combination IL-1β-CRP trajectories together was not associated with SGA status. However, women in the high trajectory IL-1β group, compared to the low, were more likely to deliver SGA infants (OR = 3.69 [1.34-10.15] P=0.011), see Table. In the low CRP trajectory, compared to high levels, SGA deliveries were increased (OR = 3.18 [1.02-9.90] P=0.046). When additionally controlling for BMI, the CRP trajectory membership was no longer significant. IL-1β and CRP levels are not useful when grouped to evaluate for SGA, however we confirm that these cytokine values are significantly correlated with SGA independently.