Purpose: To validate known and determine new predictors of non-response to ranibizumab in patients with neovascular age-related macular degeneration (AMD) and to incorporate these factors into a prediction rule.
Methods: This multicenter, observational cohort study included 391 patients treated with ranibizumab for neovascular AMD. We performed genetic analysis for single nucleotide polymorphisms in AMD-associated genes and collected questionnaires regarding environmental factors and disease history. The primary outcome was non-response to treatment, defined as a loss of visual acuity ≥30% of letters.
Results: Of the 391 patients, 47 were classified as non-responsive. Independent predictors for non-response were age, baseline visual acuity, diabetes mellitus and accumulation of risk alleles in the CFH, ARMS2 and VEGF-A genes. The area under the receiver operating characteristic curve was 0.77 (95% confidence interval 0.70-0.84). We derived a clinical prediction rule, with possible total risk scores ranging from 0-19 points. The absolute risk of non-response varied from 3-52% between risk score groups.
Conclusion: This is an important step towards a clinical prediction rule that can aid clinicians in identifying AMD patients with increased likelihood of non-response, and consequently contribute to making shared treatment decisions.
Keywords: Age-related macular degeneration; anti-VEGF treatment; non-response; prediction model; response prediction.