Thyroid Autoimmunity Does Not Delineate a Cardiometabolic or Androgenic Phenotype in Women With Polycystic Ovary Syndrome: A Pre-Specified Cross-Sectional Analysis

BackgroundThyroid autoimmunity (TAI) is frequently reported in women with polycystic ovary syndrome (PCOS), yet its clinical relevance for cardiometabolic and androgenic severity remains uncertain. We evaluated whether TAI identifies a metabolically or androgenically more severe PCOS phenotype using pre-specified exposure definitions and cardiometabolic endpoints. MethodsThis cross-sectional study included 1,300 women with confirmed PCOS in the source dataset. Thyroid autoimmunity was defined a priori using three definitions: anti-thyroid peroxidase antibodies above the laboratory upper limit of normal (TAI_A, primary definition), anti-TPO positivity combined with thyroid-stimulating hormone >4.0 mIU/L (TAI_B), and high-titer anti-TPO >100 IU/mL (TAI_C). The primary endpoint was triglyceride-to-high-density lipoprotein cholesterol ratio (TG/HDL-C) >3.5. Secondary endpoints included non-HDL-C [≥]130 mg/dL and 120-minute oral glucose tolerance test (OGTT) glucose [≥]140 mg/dL. Associations were assessed using age-adjusted Firth logistic regression models in complete-case cohorts. Sensitivity analyses included restriction to euthyroid participants, alternative TAI definitions, trimming of extreme values (1-99%), and bootstrap-based confidence intervals. Exploratory hormonal comparisons were adjusted using the Benjamini-Hochberg false discovery rate. ResultsTAI_A was not significantly associated with the primary endpoint (TG/HDL >3.5) (OR 0.77, 95% CI 0.21-1.67). No significant associations were observed for secondary endpoints including non-HDL-C [≥]130 mg/dL (OR 1.09, 95% CI 0.61-1.76) or impaired glucose tolerance on OGTT (OR 1.27, 95% CI 0.63-2.18). Results remained directionally consistent across alternative TAI definitions and sensitivity analyses, including restriction to euthyroid women and trimming of extreme values. In exploratory analyses, thyroid-stimulating hormone levels differed between TAI-positive and TAI-negative women, while no androgenic or cardiometabolic parameters remained significant after false discovery rate correction. Model diagnostics did not indicate major violations of model assumptions. ConclusionIn this large cross-sectional cohort of women with PCOS, thyroid autoimmunity was not associated with an adverse cardiometabolic or androgenic phenotype. Anti-TPO positivity alone therefore does not appear to identify a metabolically high-risk PCOS subgroup under the studied conditions. Prospective studies are needed to clarify the longitudinal implications of thyroid autoimmunity in PCOS.