Aqueous proteins help predict the response of patients with neovascular age-related macular degeneration to anti-VEGF therapy

Abstract

BackgroundTo reduce the treatment burden for patients with neovascular age-related macular degeneration (nvAMD), emerging therapies targeting vascular endothelial growth factor (VEGF) are being designed to extend the interval between treatments, thereby minimizing the number of intraocular injections. However, which patients will benefit from longer-acting agents is not clear.MethodsEyes with nvAMD (n = 122) underwent 3 consecutive monthly injections with currently available anti-VEGF therapies, followed by a treat-and-extend protocol. Patients who remained quiescent 12 weeks from their prior treatment entered a treatment pause and were switched to pro re nata (PRN) treatment (based on vision, clinical exam, and/or imaging studies). Proteomic analysis was performed on aqueous fluid to identify proteins that correlate with patients' response to treatment.ResultsAt the end of 1 year, 38 of 122 eyes (31%) entered a treatment pause (≥30 weeks). Conversely, 21 of 122 eyes (17%) failed extension and required monthly treatment at the end of year 1. Proteomic analysis of aqueous fluid identified proteins that correlated with patients' response to treatment, including proteins previously implicated in AMD pathogenesis. Interestingly, apolipoprotein-B100 (ApoB100), a principal component of drusen implicated in the progression of nonneovascular AMD, was increased in treated patients who required less frequent injections. ApoB100 expression was higher in AMD eyes compared with controls but was lower in eyes that develop choroidal neovascularization (CNV), consistent with a protective role. Accordingly, mice overexpressing ApoB100 were partially protected from laser-induced CNV.FundingThis work was supported by the National Eye Institute, National Institutes of Health grants R01EY029750, R01EY025705, and R01 EY27961; the Research to Prevent Blindness, Inc.; the Alcon Research Institute; and Johns Hopkins University through the Robert Bond Welch and Branna and Irving Sisenwein professorships in ophthalmology.ConclusionAqueous biomarkers could help identify patients with nvAMD who may not require or benefit from long-term treatment with anti-VEGF therapy.

Keywords: Clinical practice; Complement; Expression profiling; Ophthalmology.

Figure 1. Heatmap comparing fluid over time for eyes of patients who required sustained anti-VEGF treatment versus those who were successfully weaned from anti-VEGF therapy by 12 months.

OCT images were obtained from all 102 eligible patients who underwent the TEP/M approach for at least 12 months. Presence of fluid on OCT was graded independently by 2 investigators for the presence of no fluid, subretinal fluid (SRF), intraretinal fluid (IRF), or both at time points 0, 1, 2, 3, 6, and 12 months after initiation of protocol. Fluid status overtime for each individual patient is graphically represented with dark blue denoting no fluid; light blue, SRF; light green, IRF; and yellow, both. Patients were grouped into 2 categories: those not weaned (requiring sustained treatment every 4–12 weeks) and those weaned off treatment. Within each group, patients were sorted by severity of fluid (none < SRF < IRF < both).

Figure 2. Aqueous levels of VEGF in TEP/M patients.

(A) Pretreatment aqueous VEGF levels (prior to first injection) for patients with increasing interval between treatments at 12 months (from subset of TEP/M patients). (B) Posttreatment aqueous VEGF levels (having received their first injection) for patients with increasing interval between treatments at 12 months (from subset of TEP/M patients). (C) Comparison of pretreatment and posttreatment aqueous VEGF levels for patients with increasing interval between treatments at 12 months (from subset of TEP/M patients). Statistical analysis was performed using Wilcoxon rank sum test. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. An x indicates that statistical analyses could not be performed due to insufficient samples.

Figure 3. Overview of patient sample collection and proteomics screen in untreated (anti-VEGF naive) and treated (those who have received anti-VEGF therapy) patients with wet AMD.

(A) Aqueous samples were collected from clinic patients with nvAMD via anterior chamber paracentesis. (B and C) Illustration of disrupted outer blood-retinal barrier in nvAMD (B) and release of VEGF and other angiogenic mediators that promote choroidal neovascularization (C). (D) Patient samples were separated into those that required monthly (q4) treatment versus those who were able to be extended to 12 or more weeks (q12+) and by anti-VEGF treatment status (treatment naive and after treatment). (E) Pooled patient aqueous samples containing 10 μg of protein were prepared. (F) Samples were digested by proteolytic enzymes. (G) Mass spectrometry was performed to analyze each sample. (H) Unique proteins were identified using the Spectronaut Proteomics System. (I) Differential protein analysis was performed comparing q4 and q12+ nvAMD patient samples with non-AMD controls and patients with (dry) nnvAMD. (J) To evaluate the molecular effects of anti-VEGF treatment in patients, we utilized principal component analysis (PCA). The samples used in the analysis include q4 and q12+ patients and their responses to anti-VEGF treatment, as well as non-AMD controls and patients with nnvAMD. The first 3 components were selected for the 3D PCA.

Figure 4. Stepwise identification and isolation of key proteins driving phenotype of patients with nvAMD in untreated and treated q12+ versus q4 groups.

(A) Flow diagram describing the process of removing proteins with similar expression levels, those with sequences that overlapped with anti-VEGF therapies, and those that were highly variable between q4 and q12+ groups to identify proteins of interest. (B) Scatter plot of identified aqueous proteins with 2-fold changes between q4 untreated patients and q12+ untreated patients. (C) Scatter plot of identified aqueous proteins with 2-fold changes between q4 treated patients and q12+ treated patients. In B and C, colored markers represent enriched biological process from the proteomics analysis; gray dots are the proteins with no enriched biological processes.

Figure 5. Comparison of expression of key proteins in nvAMD versus nnvAMD and identification of overlapping proteins.

(A) Scatter plot of identified aqueous proteins with 2-fold changes between patients with (dry) nnvAMD and those with (wet) nvAMD. Colored markers represent enriched biological process from the proteomics analysis; gray dots are the proteins with no enriched biological processes. (B) Venn diagram describing overlapping proteins identified in the comparisons between q4 versus q12 untreated patients, q4 versus q12 treated patients, and patients with nnvAMD versus those with nvAMD (see Supplemental Table 12).

Figure 6. Comparison of expression of key immunomodulatory proteins and complement proteins.

(A) Comparison of the expression of aqueous proteins identified in the proteomics analyses between q4 versus q12 untreated patients, q4 versus q12 treated patients, and patients with nnvAMD versus those with nvAMD, highlighting immunomodulatory proteins. Proteins increased or decreased less than 2-fold were excluded from this analysis. (B) Comparison of the expression of aqueous proteins identified in the proteomics analyses between q4 versus q12 untreated patients, q4 versus q12 treated patients, and patients with nnvAMD versus those with nvAMD,s highlighting complement proteins.

Figure 7. Key proteins identified by proteomics analyses and their response to anti-VEGF treatment in q4 versus q12 patients.

(A) Proteins that increased following treatment with anti-VEGF therapy in both q4 and q12+ patients with nvAMD. (B) Proteins that decreased following treatment with anti-VEGF therapy in both q4 and q12+ patients with nvAMD. (C) Proteins that increased following treatment with anti-VEGF therapy in q4 patients with nvAMD but decreased following treatment in the q12+ patients with nvAMD. (D) Proteins that decreased following treatment with anti-VEGF therapy in q4 patients with nvAMD but increased following treatment in the q12+ patients with nvAMD. Red box highlights proteins present in all 3 comparisons (q4 vs. q12+ untreated; q4 vs. q12+ treated, and nvAMD vs. nnvAMD; see Figure 5B).

Figure 8. Apolipoprotein B-100 plays a protective role in the development of CNV.

(A) ApoB100 protein level in aqueous of non-AMD control, nnvAMD, and nvAMD. Compared with that of non-AMD control, ApoB100 was significantly higher in the aqueous of patients with nnvAMD and nvAMD. Moreover, ApoB100 levels in the aqueous of patients with nnvAMD was dramatically higher than that in patients with nvAMD. (B) ApoB100 mRNA level in neurosensory retinal, RPE/choroid, and liver of both WT and ApoB mutant mice. ApoB100 levels in all 3 tissues of ApoB mutant mice are significantly higher than those from WT mice. (C and D) Laser CNV lesion size in 9-month-old (C) and 3-month-old (D) WT and ApoB100 mutant mice 7 days after treatment with laser. Scale bars: 100 μm. Individual dots represent CNV spots from choroids of 4 mice in each case. A reduction in choroidal neovascularization was observed in ApoB100 transgenic mice compared with WT mice. Results were plotted as mean ± SD. Unpaired Student’s t test was used to compare the 2 groups. *P < 0.05. (E) Vegf mRNA level in RPE/choroid from WT and ApoB mutant mice. Results were plotted as mean ± SD. P values were generated by 2-tailed Student’s t test. (F) Laser CNV comparison between WT and ApoB mutant mice 7 days following treatment with laser. A subset of mice were treated with 200 ng aflibercept on day 3 following laser treatment. n = 4 to 8 animals for each condition. Results were plotted as mean ± SD. One-way ANOVA with Tukey’s multiple comparison test was used to compare the different treatments with each other. ****P < 0.0001; ***P < 0.001; **P < 0.01; *P < 0.05; and NS, P > 0.05.