Abstract
Purpose :
Refractive errors (RE) are strong risk factors for primary open angle glaucoma (POAG) and primary angle closure glaucoma (PACG). The causal direction of the association between RE and POAG, however, is unclear. We conducted a series of regression analyses in the UK Biobank to dissect the relationship between RE, POAG, intraocular pressure (IOP), and glaucoma medication (GRx) use.
Methods :
The UK Biobank study conducted full eye exams on ~130K individuals. The exams included autorefraction and tonometry. History of glaucoma and medication use was obtained via a touchscreen questionnaire and ICD codes were obtained through linked medical records. We estimated the relationship between all reported medications on RE. We further performed a series of regression analyses to infer causal relationships between RE, IOP, PACG, POAG, and GRx. Logistic regression was used for binary traits and linear regression was performed for quantitative traits.
Results :
There were 2,987 inferred POAG cases and 116 with PACG. 720 unique medications were used by at least 100 individuals. All but one of 14 GRx were associated with RE at p<0.05. All GRx were associated with a more myopic RE (range -0.51D to -1.80D in univariate analyses). RE was inversely associated with POAG (β=-0.38D; p=2x10-16). This association decreased after adjusting for IOP (-0.19; p=0.00023) and reversed after adjusting for IOP + GRx (β=0.15; p=0.025). PACG was associated with greater hyperopia RE (β=2.02D; p<1.2x10-15). This effect increased after adjusting for IOP (β=2.33, p<2x10-16), and IOP + GRx (β=2.50, p<2x10-16). More myopia yielded greater odds of POAG (OR=1.049/D, p=1.9x10-14). This association decreased after adjusting for IOP (OR=1.03/D, p=3x10-6) and vanished after IOP + GRx adjustment (OR=0.99/D myopia, p=0.15). Conversely, the odds of PACG and hyperopia were consistent before (OR=1.39/D hyperopia) and after adjustment for IOP (1.43/D hyperopia) and IOP + GRx (1.41/D).
Conclusions :
IOP-lowering drugs are associated with higher degrees of myopia in the general population, with effect sizes ranging from -0.51D to -1,81D. Our conditional analyses suggest a mediating effect of IOP between POAG and RE but not between PACG and RE. These results are consistent with the known pathophysiology of PACG. Although a mediating effect of IOP between RE and POAG is likely, our results do not distinguish the causal direction of this effect.
This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.