Thyroid function and age-related macular degeneration: a prospective population-based cohort study - the Rotterdam Study

In animal models, lack of thyroid hormone is associated with cone photoreceptor preservation, while administration of high doses of active thyroid hormone leads to deterioration. The association between thyroid function and age-related macular degeneration (AMD) has not been investigated in the general population.

Participants of age ≥55 years from the Rotterdam Study with thyroid-stimulating hormone (TSH) and/or free thyroxine (FT4) measurements and AMD assessment were included. We conducted age- and sex-adjusted Cox proportional hazards models to explore the association of TSH or FT4 with AMD, in the full range and in those with TSH (0.4-4.0 mIU/L) and/or FT4 in normal range (11–25 pmol/L). Cox proportional hazards models were performed for the association of TSH or FT4 with retinal pigment alterations (RPA), as an early marker of retinal changes. Multivariable models additionally included cardiovascular risk factors and thyroid peroxidase antibodies positivity. We also performed stratification by age and sex. A bidirectional look-up in genome-wide association study (GWAS) data for thyroid parameters and AMD was performed. Single nucleotide polymorphisms (SNPs) that are significantly associated with both phenotypes were identified.

We included 5,573 participants with a median follow-up of 6.9 years (interquartile range 4.4-10.8 years). During follow-up 805 people developed AMD. TSH levels were not associated with increased risk of AMD. Within normal range of FT4, participants in the highest FT4 quintile had a 1.34-fold increased risk of developing AMD, compared to individuals in the middle group (95% confidence interval [CI] 1.07-1.66). Higher FT4 values in the full range were associated with a higher risk of AMD (hazard ratio 1.04, CI, 1.01-1.06 per 1 pmol/L increase). Higher FT4 levels were similarly associated with a higher risk of RPA. Restricting analyses to euthyroid individuals, additional multivariable models, and stratification did not change estimates. We found a SNP (rs943080) in the VEGF-A gene, associated with AMD, to be significant in the TSH GWAS (P = 1.2 x 10⁻⁴). Adding this SNP to multivariable models did not change estimates.

Higher FT4 values are associated with increased risk of AMD - even in euthyroid individuals - and increased risk of RPA. Our data suggest an important role of thyroid hormone in pathways leading to AMD.