Abstract
Background: The gut microbiome is altered in obese adolescents with polycystic ovary syndrome (PCOS), and is associated with free testosterone, metabolic markers, and insulin resistance. Combined oral contraceptives (OCP) are a primary treatment for PCOS and decrease testosterone, but the effect on the serum metabolome or gut microbiome in obese adolescents with PCOS is unknown. Objective: Contrast gut microbiome profiles, targeted serum metabolomics, hormone levels, and metabolic measures in adolescents with PCOS and obesity with and without OCP treatment. Methods: Adolescent girls with obesity and PCOS underwent stool and fasting blood collection and MRI for hepatic fat fraction. Fecal bacteria were profiled by high-throughput 16S rRNA gene sequencing and fasting serum metabolomics performed with high resolution mass spectrometry. Groups were contrasted using t-tests and principle least squares discrimination analysis (PLS-DA). Associations between bacterial taxa, baseline metabolic measures, hormone levels and the metabolome were conducted using Spearman analysis. Analyses were adjusted for false discovery rate. Results: 29 adolescents with obesity [Untreated N = 21, 16 ± 1.2 years, BMI%ile 36.5 ± 3.0; OCP N = 8, 15.5 ± 0.9 years, BMI%ile 32.5 ± 3.9] participated. Of the untreated adolescents, N = 14 contributed serum for metabolomic analysis. Participants on OCP therapy had lower free testosterone and free androgen index (p < 0.001) and higher sex hormone binding globulin. There was no difference in measures of fasting glucose, insulin, lipids or HOMA-IR between groups. PLS-DA of serum metabolomics showed discrimination between the groups, secondary amino acid changes. Untreated and OCP had similar stool microbiome α-diversity based on evenness (p = 0.28), richness (p = 0.39), and Shannon diversity (p = 0.24) and global microbial composition (β-diversity, p = 0.56). There were no differences in % relative abundance at any level. Bacterial α-diversity was negatively associated with serum long chain fatty acids and branched chain amino acids. A higher %relative abundance of family Ruminococcaceae was significantly associated with serum bile acids and HOMA-IR. Conclusion: Despite hormone and serum amino acid differences, girls treated with OCP had similar metabolic and gut microbiome profiles compared to the untreated PCOS group. The association between bacterial α-diversity, Ruminococcaceae, clinical markers and long chain fatty acids suggests a potential role of the gut microbiome in the pathogenesis of the metabolic comorbidities in PCOS.