Pharmacogenetic associations with vascular endothelial growth factor inhibition in participants with neovascular age-related macular degeneration in the IVAN Study.

PURPOSE

To determine if prespecified genetic polymorphisms influence responsiveness to vascular endothelial growth factor (VEGF) inhibition in neovascular age-related macular degeneration (nAMD). The objectives were to replicate 3 reported pharmacogenetic associations of response in nAMD and to test for novel associations.

DESIGN

Cohort study, combining information about patients' genotypes with information from a randomized controlled trial about responsiveness to anti-VEGF therapy for nAMD.

PARTICIPANTS

Five hundred nine participants with nAMD, enrolled in the Alternative Treatments to Inhibit VEGF in Patients with Age-Related Choroidal Neovascularisation (IVAN) trial.

METHODS

Participants were classified as responders or nonresponders to VEGF inhibition based on the optical coherence tomography (OCT) metric of total retinal thickness (TRT). We computed the change in TRT from baseline to the latest time point for which OCT data were available (3, 6, 9, or 12 months). Eyes with changes in TRT greater than or equal to the 75th percentile or more were classified as responders, and those with changes less than or equal to the 25th percentile or lower were classified as non-responders. Three previously reported associations of response to VEGF inhibition in nAMD involving single nucleotide polymorphisms (SNPs) at the CFH, FZD4, and HTRA1/ARMS2 loci were tested for replication. An additional 482 SNPs also were tested using a candidate gene approach. Associations were estimated as odds ratios (ORs) with confidence intervals (CIs).

MAIN OUTCOME MEASURES

The primary outcome was evidence of a genetic association with response to VEGF inhibition as measured by change in TRT.

RESULTS

One hundred twenty-six participants were classified as responders and 128 were classified as nonresponders. The SNP rs10490924 in HTRA1/ARMS2 showed a borderline association with responsiveness after Bonferroni correction (OR, 1.53; CI, 0.99-2.36; P = 0.055, Bonferroni correction). None of the other 484 additional SNPs tested for association was significant after Bonferroni correction for multiple testing. The smallest corrected P value was 0.84 (P = 0.002, uncorrected) for rs9679290 in the EPAS1 (HIF2A) gene on chromosome 2. Four of the 10 most significant results were in this gene.

CONCLUSIONS

We estimated pharmacogenetic associations using high-quality phenotype data from a randomized controlled clinical trial of nAMD. No significant association or replication of previous associations were observed. Further investigation of the EPAS1 (HIF2A) gene, however, may, be merited.