Progression of Age-Related Macular Degeneration Among Individuals Homozygous for Risk Alleles on Chromosome 1 (CFH-CFHR5) or Chromosome 10 (ARMS2/HTRA1) or Both

Key Points Question What are the independent and combined associations of the 2 most common genetic risk loci for age-related macular degeneration (AMD)—chromosomes 1 (CFH-CFHR5) and 10 (ARMS2/HTRA1)—with disease progression? Findings This case series study of 502 individuals found that, after adjusting for age and baseline AMD severity, compared with individuals with 2 risk alleles exclusively at CFH-CFHR5, the time to conversion to late-stage disease was shortest among carriers of 2 risk alleles at each of the ARMS2/HTRA1 and CFH-CFHR5 loci, followed by individuals with 2 risk alleles exclusively at ARMS2/HTRA1. Meaning These results suggest that the differential association of the 2 major genetic risk loci with disease progression may be substantial and may warrant consideration in AMD clinical research.

Chr1&10-risk group). Subjects had to be ≥ 50 years of age at the first diagnosis of AMD to be included in the analysis. Eyes were excluded if any manifestation of additional retinal disease (e.g. branch retinal vein occlusion, retinal detachment) was present or if eyes had undergone any retinal surgery and/or laser treatment. Visits prior to the data such events manifested were included.

Statistical Analyses
For survival analyses of risk of progression to late-stage AMD, we defined for each eye the first visit as the earliest visit available with imaging-based documentation of a clinical phenotype of AMD (as defined above). The timepoint of conversion was defined as the first occurrence of either late-stage atrophic or late-stage exudative AMD (as defined above). Conversion to atrophic and neovascular AMD were also evaluated separately.
For these sub-group analyses, only eyes with conversion to the relevant phenotype (either atrophic or neovascular AMD were defined as events, while conversion to the other phenotype was censored. For all analyses, eyes were censored if no conversion had occurred at the last available visit. Further, eyes were censored at the timepoint of any potential confounding event. These included occurrence of other retinal diseases (e.g. branch retinal vein occlusion) or retinal surgery (e.g. retinal detachment surgery).
Survival analyses were performed using R and packages survival, survminer and coxme). [1][2][3][4] Analyses were carried out at both the eye and patient level (using both eyes of one subject as random effects) using Cox proportional hazard models. Log-likelihood statistics were used to determine if co-variates including gender andas assessed at the first visitage and AMD grade should be included. For the survival analyses at the eye level, the hazard ratio for only one eye per patient was used. If both eyes of a subject were at risk (both eyes had either early or intermediate AMD at the first visit), the following approach was defined for selecting the eye to be used: (1) the eye that converted at the same timepoint, the one with the more severe AMD grade at the first visit. The first occurrence of a loss of VA ≥ 2 lines, ≥ 3 lines and to 20/200 or worse, as compared to the first visit in eyes at risk (i.e. either early or intermediate AMD at the first visit) was compared for the Chr1&10-risk and Chr10-risk groups to the Chr1-risk, respectively. We evaluated VA changes in the better seeing eye (as assessed as the first visit) if both eyes were at risk for conversion. Snellen VA was converted into logMAR VA. Similar to the analyses of the risk of conversion into late-stage AMD, Cox proportional hazard models with log-likelihood statistics for assessment of potential covariates were applied. The significance level for all statistical tests was set at = 0.05.
All p-values generated are two-sided. When necessary, p-values were adjusted for multiple testing using a Bonferroni correction.
We assessed possible uncertainties in the exact conversion date (caused by unequally distributed times between visits) by introducing the parameter Tcon, defined as the time interval between the last recorded visit prior to conversion to late-stage AMD and the date of the first recorded conversion, which was calculated for each eye with a recorded conversion. As Tcon tends to zero, the recorded date of conversion tends to the "true" conversion date. The distribution of this parameter did not differ significantly between genetic groups when considering Tcon ≤ 5 years or Tcon ≤ 12 months (p > 0.15). We also assessed possible uncertainties in the exact conversion date by examining the total atrophic lesion size on the conversion visit. Differences between genetic groups were determined using linear mixed-effect models to account for measurements made in both eyes.

eResults. Demographic and Phenotypic Characteristics
Subjects at risk (i.e. one or two eyes with early/intermediate AMD) were younger in the Chr1&10-risk group (n = 56, median 69.3 years) at the first recorded visit as compared to those in the Chr1-risk (n = 257, 74.9 years) and Chr10-risk (n -58, 74.3 years, p < 0.001) groups ( Table 1) Over time, a comparatively larger number of conversions was observed in both the Chr1&10-risk and Chr10-risk groups; this was driven mainly by conversions to neovascular disease, as compared to atrophic AMD (eTable 2). The age at time of conversion was similar between the Chr1-risk (80.6 years) and Chr10-risk (82.6 years) genetic groups. However, it was markedly younger in the Chr1&10-risk group (73.6 years) (p < 0.001). Out of the 394 eyes at risk of conversion at the first visit, 34 (8.6%) were lost to follow-up because of death. Of these 34 eyes, there were 26 (9.7%) in the

eFigure 1. Survival Curves for Conversion Data Based Upon Tcon ≤5 Years
Survival curves for conversion to any late-stage AMD phenotype (left), atrophic late-stage (middle) and neovascular latestage (right) in the Chr1-risk (blue), Chr10-risk (orange) and Chr1&10-risk (green) groups, shown for the analyses including one eye per subject and adjusted for age at first visit and refined AMD grade at first visit. Data are shown for persons with 5 or fewer years between the last visit before the conversion visit and the actual visit of conversion (Tcon ≤ 5). Chr1-risk indicates homozygous for risk variants at chromosome 1 without risk at Chr10; Chr10-risk, homozygous for risk variants at chromosome 10 without risk at Chr1; and Chr1&10-risk, homozygous for risk variants at chromosomes 1 and 10.

. eFigure 2. Survival Curves for Loss of Visual Acuity Based Upon Tcon ≤ 5 Years
Survival curves for loss of visual acuity in the Chr1-risk (blue), Chr10-risk (orange) and Chr1&10-risk (green) groups, shown for loss ≥ 2 lines (left), ≥ 3 lines (middle) and to 20/200 (right). The graphs are adjusted for age at first visit, refined AMD grade at first visit and visual acuity at first visit. Data are shown for persons with 5 or fewer years between the last visit before the conversion visit and the actual visit of conversion (Tcon ≤ 5). Chr1-risk indicates homozygous for risk variants at chromosome 1 without risk at Chr10; Chr10-risk, homozygous for risk variants at chromosome 10 without risk at Chr1; and Chr1&10-risk, homozygous for risk variants at chromosomes 1 and 10. Abbreviations: Chr1-risk, homozygous for risk variants at chromosome 1 without risk at chromosome 10; Chr10-risk, homozygous for risk variants at chromosome 10 without risk at chromosome 1; Chr1&10, homozygous for risk variants at chromosomes 1 and 10; iRORA, incomplete retinal pigment epithelium and outer retinal atrophy; PED, pigment epithelium detachment; VA = visual acuity; 95% confidence intervals (CI) are shown in brackets, following hazards ratios and data refers to persons with 12 or fewer months between the last visit before the conversion visit and the actual visit of conversion.