342 - Impact of Maternal Cardiometabolic Risk Factors on Human Milk Metabolomic Profiles among Mothers with Obesity

Publication Number: 3329.342

Margaret L. Ong, Brigham and Women's Hospital, Boston, MA, United States; Carmen Monthe-Dreze, Women & Infants Hospital of Rhode Island, Providence, RI, United States; Chloe Andrews, Women & Infants Hospital of Rhode Island, Wrentham, MA, United States; Patricia A. Sheridan, The Ohio State University Wexner Medical Center, Chapel Hill, NC, United States; Camilia R. Martin, Weill Cornell Medicine, New York, NY, United States; Sarbattama Sen, Women & Infants Hospital of Rhode Island, providence, RI, United States

Background: We have previously found associations between the metabolomic profile of human milk (HM) and infant growth and neurodevelopmental outcomes. However, the role of maternal metabolic health on HM composition has been understudied.

Objective: To investigate associations between maternal cardiometabolic risk factors and HM metabolomic profiles of women with obesity.

Design/Methods: This is an observational sub-study (n=26) of a randomized controlled trial (NCT02802566) of a BMI-based prenatal vitamin in women with obesity. HM samples were self-collected from a first-morning full-breast expression at 5.5±2.0 weeks postpartum. We examined the following maternal cardiometabolic risk factors: gestational glucose intolerance (GGI, glucose challenge test ≥135 mg/dL), hypertension diagnosis during pregnancy or postpartum (HTN), and serum C-reactive protein (CRP, mg/L) at initial visit. The HM metabolome was analyzed via untargeted ultra-performance liquid chromatography with tandem mass spectrometry. We used principal component analysis to consolidate 669 metabolites into factors and retained factors using scree plots and eigenvalue >1. We used linear regression models to assess associations between maternal cardiometabolic risk factors and factor scores, in unadjusted (Model 0) and adjusted models including the other risk factors without (Model 1) and with study group (Model 2). We performed Over-Representation Analysis (ORA) for the top loading metabolites into significantly associated factors.

Results: Participants delivered at mean±SD 38.6±2.0 weeks' gestation. Mothers had mean±SD age of 32.8±5.1 years, BMI 35.7±4.4 kg/m2, and CRP 9.1±11.2 mg/L at first study visit. GGI was present in 35% (n=9) and HTN in 19% (n=5) of mothers. CRP was associated with higher Factor 1 score (Table 1, ꞵ [95% CI]: 0.04 [ 0.004, 0.07]) in Model 0 and Model 1. This association was slightly attenuated in Model 2 (0.04 [-0.001, 0.07]). Metabolites that loaded negatively into Factor 1 involved pyrimidine metabolism and fatty acid biosynthesis pathways (Figure 1, Table 2). ORA for positively loading metabolites was not significant. Neither GGI nor HTN were associated with HM metabolite factors.

Conclusion(s): In women with obesity, CRP was associated with Factor 1, which included metabolites involved in pyrimidine and fatty acid metabolism. This suggests that maternal inflammation may play a role in HM composition, impacting pathways involved in child growth and development. Further research should investigate the mediating role of inflammation on HM composition and its influence on child health in larger cohorts of mothers with obesity.

Table 1: Associations between Maternal Cardiometabolic Risk Factors and Human Milk Metabolite Factor Scores
CRP: C-reactive protein. Linear regression model results are reported as beta estimates (95%) indicating change in factor score for presence of gestational glucose intolerance (reference: no gestational glucose intolerance), presence of hypertension (reference: no hypertension), for each 1 unit increase in CRP value, or study group (reference: group A). In the unadjusted analysis, maternal cardiometabolic risk factors were assessed as individual exposures for association with factor scores. In the adjusted analysis, all cardiometabolic risk factors were included as exposures to assess for independent associations with factor scores, adjusting for all other exposures in the model. Bolded shaded boxes represent findings for which p<0.05.

Figure 1: Over-Representation Enrichment Analysis for Factor 1
Enrichment ratio: # hits / # expected hits for specific metabolic pathways. The top 20 identifiable metabolites for Factor 1 were used for Metabolite Set Enrichment Analysis / Over-Representation Analysis in MetaboAnalyst 6.0 (https://www.metaboanalyst.ca/). Enrichment analyses were performed separately for a) 11/20 negatively loading metabolites, and b) 9/20 positively loading metabolites. Enrichment analysis for negatively loading metabolites was significant for pyrimidine metabolism and fatty acid biosynthesis (p < 0.05, q > 0.05). Enrichment analysis for positively loading metabolites was not significant (p >0.05). Pathway enrichment with -log10 (p-value) > 1.3 (p-value < 0.05) is noted by the red-dashed line.

Table 2: Metabolite Composition (Top 20 Identifiable) of Factor 1