A total of 65 individuals met the inclusion/exclusion criteria and were enrolled in the study. Sixty-four patients completed the study (primary end point: 12 months). One participant died of a cause unrelated to the study. Enrollment took almost 1 year due to the limited number of treatment-naïve, new-onset neovascular AMD patients presenting to the participant sites, together with other studies actively enrolling neovascular AMD patients.
All participants had new-onset neovascular AMD in the study eye. None of the study eyes had previously received treatment (thermal laser, PDT, intravitreal therapy) for neovascular AMD. In each case, vision loss was determined by the investigators, at presentation, to have occurred as a result of neovascular AMD, and no other etiology was evident. Only one eye per patient was enrolled in the study.
The fluorescein angiographic lesion characteristics of study eyes were as follows: occult without classic (21, 32%), minimally classic (21, 32%), predominantly classic (19, 29%), and classic (4, 7%) CNV. At the end of the study, 38% of individuals had no angiographic or OCT evidence of persistent leakage or active CNV.
Independent of genotype, individuals in the study presented with a mean best-corrected visual acuity of 60.23 ETDRS letters (SD ±15.4). During the course of the study, visual acuity improved by 3.84 (±9.02) letters at 6 months and by 5.8 letters (±9.6) at 12 months. Mean baseline central macular thickness on OCT was 329.74 (±89.74) μm at baseline and decreased by a mean of 107.62 (±115.54) μm at 6 months and 92.26 (±140.96) μm at 12 months. The mean number of injections required was 6.19 (±3.26). CRP level at baseline was 0.31 mg/L (±0.27).
STUDY INTERVENTION AND ADVERSE EVENTS
LGS participants received a median number of six injections during the study (not including the initial injection at baseline). No injections were declined. No other therapies were required, and there were no reports from treating physicians of deviations from the re-treatment criteria.
Since patients were receiving on-label therapy with ranibizumab, strictly, the “intervention” in this study was venepuncture. There were no adverse events relating to this procedure. One patient died during the study period of a cause unrelated to his ocular status or treatment. No other adverse events (Grade 2, NCI grading system or greater) were recorded. Other mild adverse events were considered unrelated to study medication or the protocol.
GENOTYPING AND QUALITY CONTROL
The Illumina (San Diego, California) 660-Quad SNP chip genotyped >550,000 SNPs in 44 individuals, and of these, 66,405 failed quality control and were excluded from further analysis (477 were out of Hardy-Weinberg equilibrium while the rest had either a minor allele frequency <0.05 or had too many failed calls).
Analysis was conducted in four phases. In Phase I, differences between the enrolled population and a similar population were investigated to determine how representative the LGS participants were of newly diagnosed neovascular AMD in general. In Phase II, a candidate gene approach was utilized to reduce the correction needed for multiple testing. This included the primary end point analysis of change in visual acuity at the end of the study (12 months). In Phase III, all SNPs in the genome scan were included. Phase IV examined all SNPs, both candidates and genome-wide association, as to whether they influenced baseline characteristics of participants.
PHASE I: VALIDATION ANALYSIS
shows summary demographic, environmental exposure, and genetic risk factor data for the individuals in the LGS. When compared with a similar hospital, academic practice population (the CEI sporadic cases with CNV), the LGS participants were similar in age (marginally older, P=0.04) and had achieved a higher education status. In all other respects, the LGS patients were very comparable. Genotypic comparisons between LGS and CEI populations were limited to CFH and ARMS2 because the minor allele frequencies in these two genes are high enough for meaningful evaluation in the small LGS cohort. There were no significant differences between allele frequencies for these two SNPs.
SUMMARY DATA FOR DEMOGRAPHIC AND PERSONAL FACTORS FOR THE INDIVIDUALS IN THE LUCENTIS GENOTYPE STUDY (LGS) AS COMPARED WITH THE CASEY EYE INSTITUTE SPORADIC CASES WITH CHOROIDAL NEOVASCULARIZATION (CEIMDC CNV)
PHASE II: TREATMENT RESPONSE ANALYSIS AMONG CANDIDATE GENES
The following end points were investigated: best-corrected ETDRS letter score in the treated eye at 6 and 12 months, change in central macular thickness at 6 and 12 months, angiographic evidence of persistent leakage at 6 and 12 months, and number of injections received by end of study. Logistic or linear regression, as appropriate, was used to evaluate associations between these measures and a list of candidate genes considered likely to play a role in angiogenesis. The candidate gene list () comprised known AMD-susceptibility genes and genes involved in the control of angiogenesis. The full list of SNPs investigated is shown in the .
SINGLE-NUCLEOTIDE POLYMORPHISMS SHOWING GREATEST ASSOCIATIONS IN GENOME-WIDE ANALYSES
The primary end point for the study was best-corrected ETDRS letter score in the treated eye at 12 months. Two genes showed statistically significant association (after correction for multiple testing) with change in visual acuity, CFH and CTGF (). Having the minor allele (A) in rs1065489 (CFH) and rs9399005 (CTGF) conferred a worse visual outcome compared with the ancestral allele. At 6 months, CFH is also significantly associated (), suggesting a consistent influence on visual improvement of variants in this gene.
CANDIDATE GENE ANALYSIS: DIFFERENCE IN BEST-CORRECTED VISUAL ACUITY AT 12 MONTHS
CANDIDATE GENE ANALYSIS: DIFFERENCE IN BEST-CORRECTED VISUAL ACUITY AT 6 MONTHS
When OCT central macular thickness is examined in the same way, different candidate genes appear significant (). The results show that the minor allele of the same SNP in the gene for complement factor 3 (rs2230205) is associated with greater reduction in retinal thickness at both 6 () and 12 months. SNPs in two other genes, thombospondin-1 (THBS1) and fibroblast growth factor receptor-2 (FGFR2), are also associated with improved treatment response to ranibizumab therapy. A SNP in FGFR2 is also associated with persistent leakage on fluorescein angiography at 12 months (), underlining the potential role of this gene in treatment response. Other genes are also noted to be associated, including SNPs in FLT1 ( and ).
CANDIDATE GENE ANALYSIS: CHANGE IN CENTRAL MACULAR THICKNESS AT 6 MONTHS
CANDIDATE GENE ANALYSIS: FLUORESCEIN ANGIOGRAPHIC EVIDENCE OF PERSISTENT LEAKAGE FROM CHOROIDAL NEOVASCULARIZATION AT 12 MONTHS
CANDIDATE GENE ANALYSIS: FLUORESCEIN ANGIOGRAPHIC EVIDENCE OF PERSISTENT LEAKAGE FROM CHOROIDAL NEOVASCULARIZATION AT 6 MONTHS
Several SNPs in three genes (CFH, VEGFA, and FLT1) are strongly associated with number of injections received during the study (). This parameter was measured only at the primary end point of 1 year. In the case of VEGFA and FLT1, possessing the minor allele of each SNP resulted in the need for fewer injections. By contrast, those with the minor allele in the CFH gene needed more injections.
CANDIDATE GENE ANALYSIS: NUMBER OF MONTHLY RANIBIZUMAB INTRAVITREAL INJECTIONS OVER 12 MONTHS
PHASE III: GENOME-WIDE TREATMENT RESPONSE ANALYSES
The following end points were investigated: best-corrected ETDRS letter score in the treated eye at 6 and 12 months, change in central macular thickness at 6 and 12 months, angiographic evidence of persistent leakage at 6 and 12 months, and number of injections received by end of study. Logistic or linear regression, as appropriate, was used to evaluate associations between these measures and all SNPs successfully genotyped on the Illumina 660-Quad SNP chip.
A conservative P value for significance of P<10−7 was employed, after which only one SNP near the gene noncoding RNA 158 (NCRNA00158), a member of the iRNA family, achieved genome-wide statistical significance (, ), the minor allele of which was associated with a much thicker central macula at 12 months than those with the ancestral allele. and are a Manhattan plot whereby the y-axis shows the P value for each SNP in order along each chromosome. When a slightly less stringent correction is applied, then several SNPs achieve significance (), including several known to be expressed in the retina. TSHZ2, CCDC102B, GRIA3, PTPRD, FLJ42392, SETD2, KCNQ5, and ME3 were similarly associated with macular thickness at 12 months as NCRNA00158. The minor allele of Protocadherin-19 (PCDH19) was associated with a substantially worse visual outcome at the 6-month interim visit.
GENOME-WIDE ASSOCIATION ANALYSIS: SINGLE-NUCLEOTIDE POLYMORPHISMS ACHIEVING GENOME-WIDE SIGNIFICANCE LEVELS AFTER BONFERRONI CORRECTION FOR MULTIPLE TESTING
Manhattan plot of genome-wide association analysis. Change in central macular thickness at 12 months.
Manhattan plot of genome-wide association analysis. C-reactive protein at baseline.
GENOME-WIDE ASSOCIATION ANALYSIS: SINGLE-NUCLEOTIDE POLYMORPHISMS ACHIEVING MARGINAL GENOME-WIDE SIGNIFICANCE LEVELS AFTER BONFERRONI CORRECTION FOR MULTIPLE TESTING*
PHASE IV: ASSOCIATIONS WITH BASELINE CHARACTERISTICS
The following baseline characteristics were evaluated: best-corrected ETDRS letter score in the treated eye, fluorescein angiographic lesion characteristics, baseline OCT central macular thickness, and CRP at baseline. When examined on a candidate gene basis, as in Phase II of the analysis, the minor alleles of SNPs in VEGFA, C3, and PF4 (platelet factor 4) were associated with a worse visual acuity at presentation (). The same SNP in C3 is also associated with having a more thickened retina at presentation (). This is the same SNP associated with less response to ranibizumab therapy throughout the study (6 and 12 months). Three SNPs in the fibroblast growth factor receptor family (FGFR-1 and -2) are also associated with worse macular thickness at presentation. FGFR2 and C3 were also associated with having a higher CRP at baseline ().
CANDIDATE GENE ANALYSIS: BASELINE BEST-CORRECTED VISUAL ACUITY (ETDRS LETTERS)
CANDIDATE GENE ANALYSIS: BASELINE CENTRAL MACULAR THICKNESS
CANDIDATE GENE ANALYSIS: SERUM C-REACTIVE PROTEIN (MG/L) AT BASELINE
Baseline CRP was also highlighted in the genome-wide analyses where SNPs in LPIN1, ABCG2, and ZNF18A were associated (). Worse baseline visual acuity was associated with SNPs in CCDC102B, RGS6, FAM76B, all three being robustly expressed in the retina. RIPK3 and SALL1 were associated with more thickened maculae on OCT. These two are also known to be expressed in the retina.