Sex-, and neurodegeneration-dependent effect modification in the association between thyroid function and cognitive impairment in non-depressed, non-demented elderly

Introduction: Understanding the particularities of thyroid-cognition interactions in the elderly is crucial in assessing the risks and evaluating therapeutic options. Methods: Cross-sectional analyses where participants from Alzheimer s Disease Neuroimaging Initiative (ADNI) with mild cognitive impairment (MCI) and healthy controls (HC), with complete neurocognitive tests, thyroid stimulating hormone (TSH) <10 uIU/mL, and geriatric depression scale (GDS) <5 were eligible. Linear and logistic regression models, including testing for non-linearity, were performed. Sex and neurodegeneration-related stratifications were explored. Results: Of the total 1845 participants, with a median age of 73 (IQR: 68, 78); 887 (48%) were females, and 1056 (57%) had MCI. The median TSH level was 1.70 uIU/mL (IQR: 1.15, 2.40). There was a significant association between TSH and cognition only in males (adj. Beta Males: -0.40; 95% CI: -0.74, -0.07; p-value: 0.019). The odds of being diagnosed with MCI at baseline decreased with higher TSH levels in the total study population (adj. OR Total: 0.87; 95% CI: 0.79, 0.95; p-value: 0.002), and in males (adj. OR Males: 0.80; 95% CI: 0.70, 0.92; p-value: 0.001). The median TSH value was a significant cutoff in this association. Conclusions: The association between thyroid function and cognitive decline in the elderly is subject to a sex-driven effect modification and depends on a cutoff value.