Cardiometabolic, thyroid, and cardiac biomarker associations with insulin resistance: cross-sectional study from a Romanian cohort

Insulin resistance (IR) is a key driver of the cardiometabolic continuum. However, its associations with thyroid dysfunction, systemic inflammation, and subclinical myocardial injury have not been jointly evaluated at population scale using adjusted analyses. This study aimed to characterize the biomarker profile of IR and to determine which associations with HOMA-IR persist after adjustment for age, sex, and the triglyceride/HDL-cholesterol (TG/HDL) ratio. We conducted a retrospective cross-sectional study using anonymized laboratory data from a national private network in Romania. Of 104,406 individuals with at least one HOMA-IR determination, 98,442 adults (≥ 18 years) formed the primary cohort. Associations between HOMA-IR and 16 biomarkers were assessed using Spearman correlations with Benjamini–Hochberg correction in paired sub-cohorts. Four pre-specified multivariable models were fitted on complete-case datasets: logistic regression for subclinical hypothyroidism ( n = 8,039) and linear regression for log(TSH) ( n = 12,370), log(hsCRP) ( n = 1,700), and log(hsTnT) ( n = 297), each with HOMA-IR as the exposure and adjustment for age, sex, and TG/HDL ratio. The derivation of every analytical cohort is set out in Table 1. At least one HOMA-IR value > 2 was recorded in 69.3% of tested individuals. HOMA-IR showed a very strong correlation with insulin (ρ = 0.969) and moderate correlations with triglycerides (ρ = 0.499), fasting glucose (ρ = 0.498), and HDL-cholesterol (ρ = −0.413). hsCRP showed a weak positive correlation (ρ = 0.346), whereas thyroid and hsTnT correlations were weak or very weak. In adjusted analyses, HOMA-IR remained associated with log(hsCRP) (β = 0.071, 95% CI 0.056–0.086, p < 0.001; model R² = 7.4%) and with log(TSH) (β = 0.009, 95% CI 0.004–0.013, p < 0.001; model R² = 0.9%). HOMA-IR was not independently associated with subclinical hypothyroidism (OR = 1.015, p = 0.253) or with log(hsTnT) (β = 0.004, p = 0.635) after adjustment. In this large retrospective laboratory cohort, HOMA-IR was associated primarily with atherogenic dyslipidemia and, after adjustment, with hsCRP. The association with TSH was statistically significant but small in magnitude and accounted for less than 1% of TSH variance. Associations with subclinical hypothyroidism and hsTnT did not persist after adjustment. These findings should be interpreted as cross-sectional associations in a referral-based population and require prospective confirmation before any screening or therapeutic implication can be inferred.