Impact of Baseline Quantitative OCT Features on Response to Risuteganib for the Treatment of Dry AMD—The Importance of Outer Retinal Integrity

Abstract

Objective:

The purpose of the study was to perform a post-hoc analysis to explore the effect of baseline anatomic characteristics identified on optical coherence tomography (OCT) on best-corrected visual acuity (BCVA) response with risuteganib from the completed Phase 2 study in subjects with dry age-related macular degeneration.

Design:

Post-hoc analysis of a randomized, double-masked, placebo-controlled Phase 2 study

Subjects, participants, and/or controls:

Eyes with intermediate dry AMD with BCVA between 20/40–20/200. Patients with concurrent vision impacting or macula obscuring ocular pathology were excluded.

Methods:

Patients were randomized to receive intravitreal 1.0 mg ristuteganib or sham injection at baseline. A second 1.0mg intravitreal injection of risuteganib was given at week 16 in the treatment arm. Two independent masked reading centers evaluated baseline anatomic characteristics on OCT to explore for features associated with positive response to risuteganib.

Main outcome measures:

Treatment response was defined as a gain of ≥ 8 letters in BCVA from baseline to week 28 in the treatment arm compared to baseline to week 12 in the sham group. Anatomic parameters measured by retinal segmentation platforms including measures of retinal thicknesses were compared between the responders and non-responders to risuteganib.

Results:

Thirty-nine patients completed the study and underwent analysis. In the treatment arm, 48% of eyes demonstrated treatment response compared to 7% in the sham group. In quantitative anatomic assessment, enhanced ellipsoid integrity, greater outer retinal thickness and decreased geographic atrophy were associated with increased BCVA gain to risuteganib.

Conclusions:

This post hoc analysis demonstrated that baseline OCT features may help determine likelihood of functional response to risuteganib. Utilization of higher order OCT feature characterization may provide an important biomarker for treatment response and could facilitate optimized clinical trial enrollment and enrichment.

Precis:

In this analysis, advanced compartmental OCT analysis and retinal feature extraction identified key biomarkers that were associated with treatment response to risuteganib. This assessment supports the potential that imaging biomarker characterization may have for personalized care, clinical trial enrichment, and targeted therapy.

INTRODUCTION

Age-related macular degeneration (AMD) is a common degenerative disease that is predicted to affect 288 million people by 2040.¹ It is the leading irreversible contributor to blindness among elderly populations. As life expectancies have generally trended upward globally, the impact of AMD on society continues to increase in terms of burdens at both personal and health resource utilization levels.² Of the two subtypes, non-neovascular, or dry, AMD (NNVAMD) is the most prevalent and includes more advanced vision-impairing disease such as geographic atrophy (GA). Although neovascular, or wet, AMD (NVAMD), has multiple proven treatments that prevent vision loss, there are no current FDA-approved therapies for NNVAMD. ^3–5

Given the absence of disease altering therapies and its significant impact in vision loss, NNVAMD remains a major focus of ongoing research for new therapeutics. With the multiple pathogenic mechanisms involved, development work is evaluating numerous pathways for disease intervention. Among these, increased oxidative stress triggered by mitochondrial bioenergetics changes associated with age are associated with perturbation of the complement system.^6–8 Drusen, a key feature of AMD, disrupt normal metabolism involving the retinal pigment epithelium and Bruch’s membrane propagating disruption to photoreceptor debris transportation. Drusen also contain factors associated with complement pathway activation which has been associated with AMD pathogenesis and choriocapillaris loss through the membrane attack complex.^9,10

Risuteganib (Allegro Ophthalmics, San Juan Capistrano, CA, USA) is a peptide that with cytoprotective, anti-inflammatory, and pro-mitochondrial effects that may help mitigate AMD-associated cellular injury.^11,12 In a phase IIa clinical trial that assessed the effect of intravitreal risuteganib in eyes with NNVAMD, there was a statistically significant visual acuity gain in drug-treated eyes compared to sham controls.¹³ The trial reported that 48% of subjects experienced an ≥ 8 letters ETDRS BCVA gain from baseline in the risuteganib group compared to 7% in controls and 20% of the risuteganib group experienced a ≥ 15 letters improvement compared to 0% in the controls. These visual acuity results were promising and further investigations are planned to validate these findings in larger cohorts.

In a developing era of precision medicine, it has become increasingly important to understand how emerging treatments can apply to different populations and individuals, especially within diseases with complex pathophysiology such as AMD. Novel optical coherence tomography (OCT) segmentation platforms have been developed to characterize imaging biomarkers at quantitative levels.^14–16 These systems can provide a detailed microstructural OCT phenotype of retinal disease such as quantitative measurements of zonal outer retina integrity.^17–19

The primary goal of this study was to investigate whether baseline OCT anatomic features and biomarkers were associated with treatment response to risuteganib in intermediate AMD eyes. Given the unique aspect of the measurements required for this study, two independent platforms from two advanced OCT reading centers were utilized to evaluate the potential impact of outer retinal features on treatment response.

METHODS

Clinical Trial Methodology

This was a post-hoc analysis of a phase IIa prospective, randomized, double-blind, multicenter trial (NTC03626636) investigating risuteganib efficacy and safety.¹³ The study was undertaken at 7 sites across the US and had site-specific approval by Institutional Review Boards. Study design adhered to the tenants of the Declaration of Helsinki. The full methods of this trial are reported at https://clinicaltrials.gov/ct2/show/NCT03626636.¹³

In brief, the clinical trial included adults aged between 50 and 85 with a diagnosis of intermediate NNVAMD defined by 1 or more large druse (>125um) and/or multiple intermediate drusen (63–124um) and/or a combination of macular RPE disturbances. To be included in this study, eyes with symptomatic intermediate AMD, defined as ETDRS BCVA of 33 to 72 letters and decline in visual acuity over prior 12 months that was driven solely by AMD. Key exclusion criteria included NVAMD, anti-vascular endothelial growth factor therapy history, coexistent ocular pathology that could affect BCVA or macula visualization, geographic atrophy affecting the central 1mm, and history of retinal surgery.

Eligible eyes were randomized into experimental treatment or sham groups. Risuteganib (1.0mg/0.05mL) was given by intravitreal injection to eyes in the drug treatment group on Day 0 and again at week 16. Masked investigators assessed imaging and visual acuity during the study. Responders were defined those eyes that had an 8 or more letter gain after treatment. Two reading centers independently evaluated the baseline imaging in a masked fashion.

Cole Eye Advanced OCT Biomarker Analysis and Imaging Feature Extraction

In a masked fashion, pretreatment SD-OCT macular volume scans (Spectralis, Heidelberg based on protocol) were uploaded into a previously described software platform that enables linear, area, and volumetric features of multiple imaging biomarkers through machine learning enhanced feature extraction.^{14,15,17,18,20} As the imaging protocol prespecified use of the Heidelberg system, any other imaging devices were excluded for this analysis. Using automated analysis, scans underwent multi-layer segmentation including internal limiting membrane (ILM), outer nuclear layer (ONL), ellipsoid zone (EZ), external limiting membrane (ELM), retinal pigment epithelium (RPE), and Bruch’s membrane (BM), that enabled assessment of various retinal compartments and sub-RPE compartments, such overall retinal thickness (i.e., ILM-RPE) and ellipsoid zone integrity (i.e., EZ-RPE) with manual correction as needed. An image analyst reviewed each B-scan for retinal layer segmentation accuracy. A secondary quality control pass was performed by a senior project lead to evaluate imaging and segmentation consistency.^18,21

After segmentation quality review, compartmental quantitative parameters of interest, including measures of ellipsoid zone (EZ)-RPE thickness (i.e., a surrogate for photoreceptor outer segment length and EZ integrity) and RPE-BM thickness (i.e., a surrogate for drusen burden), were exported. The en face EZ-RPE topographic map engendered evaluation of panmacular percentage of total EZ attenuation (EZ-RPE thickness = 0 μm), partial EZ attenuation (EZ-RPE thickness ≤ 20 μm) and total RPE attenuation (RPE-BM thickness=0; i.e., GA). Parameters were exported in specific regions of interest, specifically the central subfield (central 1 mm), central macular subfield (central 2 mm), and panmacular outputs.

Duke OCT Advanced OCT Biomarker Analysis and Imaging Feature Extraction

Retinal boundaries and layers on study OCT images were segmented automatically with in-house custom developed software.^22–24 OCT segmentation of was performed on all images. For the analysis, ILM, outer plexiform layer (OPL)/ONL boundary, ELM, photoreceptor outer segment tip/inner retinal pigment epithelial (OS/RPE) boundary, and outer RPE/BM were segmented. Mean retinal thickness and volumes of 5 different layers were calculated on all images for the foveal center point and for the foveal central subfield and the inner and outer rings of a standard ETDRS grid, as described below:

Total retinal thickness was calculated as the distance from the innermost hyper-reflective line corresponding to the ILM to the outermost hyper-reflective line boundary corresponding to the outer RPE/BM boundary. The inner retina was calculated from the ILM to the OPL/ONL boundary. The outer retina was calculated from the OPL/ONL boundary to the OS/RPE boundary. The photoreceptor layer was calculated from the ELM to the OS/RPE boundary; 5) the RPE drusen complex (RPEDC) layer was calculated from the OS/RPE boundary to the outer RPE/BM boundary.

A previously reported novel deep learning-based algorithm, Deep OCT Atrophy Detection was utilized to automatically segment EZ defect areas through classification of 3-dimensional A-scan clusters as normal vs. defective using a convolutional neural network.^25,26 OCT images from subjects at baseline were analyzed using this algorithm.

In addition to the quantitative analysis of OCT images, the masked readers at the Duke Reading Center performed qualitative assessment of the OCT images at baseline to identify geographic atrophy (GA) anywhere in the retina, in the fovea (1-mm central subfield), and in the foveal center.

Statistical Analysis

Assessments were made independently from both the Cole and Duke reading centers using Two Sample t-tests if the variables were considered to have a normal distribution based on Shapiro-Wilk testing or Wilcoxon Rank-Sum if the distributions were non-normal. Within the group that received risuteganib, responders were compared to nonresponders. Additionally, as a control comparison, baseline features were compared between the risuteganib and sham groups. When ratios were compared (e.g. percentage of GA in a group), a test of two proportions was used. A p-value below 0.05 was considered significant. All statistical analyses were conducted using R (R Core Team).

RESULTS

Subject Clinical Characteristics and Visual Outcomes

This phase 2a exploratory study¹³ enrolled 45 subjects that were randomized into 2 groups: 29 in the treatment arm and 16 in the sham arm. Exclusions were made based on development of exclusion criteria (n=1), subject withdrawal (n=1), inability to follow instructions appropriately regarding BCVA (n=1), and patients that did not receive all assigned treatments due to adverse events (n=3). Demographic characteristics between groups are reported in Table 1 with no differences between treatment arms. The primary endpoint of gaining 8 or more ETDRS letters at 28 weeks was achieved in 12 of 25 subjects (48%, responders) who received risuteganib and 1 of 14 subjects (7.1%) who received the sham injection (p=0.013).

Table 1.

Demographic characteristics of study participants

Demographic characteristic	Sham (n=14)	Risuteganib (n=25)	P-value
Male	3 (21.4%)	9 (36.0%)	0.48
Female	11 (78.6%)	16 (64.0%)	-
Caucasian	14 (100%)	24 (96.0%)	1
Asian	0	1 (4.0%)	-
Age (mean + SD)	78.8 (8.4)	75.9 (8.43)	0.32

Cole Eye Reading Center Imaging Biomarker Analysis

Anatomic Measurements at Baseline Based on Treatment Arm

At baseline, there were no significant difference in retinal compartment parameters between the sham group and the treatment group (p>0.5). The average retinal CST was 236.38 μm in the treatment arm vs. 238.31 μm in the sham arm (p=0.83), and the mean EZ-RPE CST was 28.72 μm vs 28.33 μm (p=0.95), respectively. The RPE-BM CST mean values were 39.02 μm in the risuteganib treated group and 42.12 μm in the sham treated group (p=0.88).

Compartmental OCT Measurements at Baseline Based on Risuteganib Responder Status

At baseline, eyes that responded to risuteganib (i.e., eyes that gained 8 letters or more) had significantly greater mean thickness in multiple retinal metrics compared with eyes that did not respond to risuteganib: mean total retinal CST (256.41 vs 218.02 μm; P=0.011), mean total retinal central macular thickness (296.12 vs 264.10 μm; P=0.011), ONL-RPE CST (151.46 vs 123.78 μm; P=0.007), mean ONL-RPE central macular thickness (133.14 vs 111.21 μm; P=0.02), and mean EZ-RPE CST (35.96 vs 22.07 μm; P=0.04) (Table 2).

Table 2.

Quantitative Anatomical Measurements at Baseline for Risuteganib Nonresponder Eyes Versus Responder Eyes

Treatment Group	Risuteganib Nonresponder (n=12)		Risuteganib Responder (n=11)		Risuteganib Treatment (n=23)		Sham Treatment (n=12)		Mann-Whitney U test (P-value)
OCT Parameter	Mean	SD	Mean	SD	Mean	SD	Mean	SD	Responders vs Non-Responders
Actual Retinal Central Subfield Mean Thickness (μm)	218.02	32.07	256.41	31.65	236.38	36.79	238.31	33.45	0.011
Actual Retinal Mid Subfield Mean Thickness (μm)	264.10	22.21	296.12	24.51	279.42	28.06	278.62	29.64	0.003
EZ-RPE Central Subfield Mean Thickness (μm)	22.07	16.73	35.96	9.02	28.72	15.08	28.33	14.54	0.044
EZ-RPE Mid Subfield Mean Thickness (μm)	25.81	15.14	37.55	4.36	31.43	12.62	28.13	13.28	0.069
ONL-RPE Central Subfield Mean Thickness (μm)	123.78	22.65	151.46	17.04	137.02	24.26	134.36	23.35	0.007
ONL-RPE Mid Subfield Mean Thickness (μm)	111.21	22.55	133.14	10.34	121.70	20.69	115.34	24.89	0.019
RPE-BM Central Subfield Mean Thickness (μm)	33.55	22.44	44.98	40.59	39.01	32.17	42.12	31.90	0.347
RPE-BM Mid Subfield Mean Thickness (μm)	28.98	18.65	38.31	24.58	33.44	21.71	37.55	37.42	0.211

Abbreviations: EZ, ellipsoid zone; RPE, retinal pigment epithelium; ONL, outer nuclear layer; BM, Bruch’s membrane

Representative comparative examples of OCT and retinal compartment integrity/thickness maps between a responder and non-responder eye are shown in Figure 1 and Figure 2. Both retinal thickness maps (ILM-RPE, Figure 1B and 2B) reveal primarily normal/similar images. However, the risuteganib responder eye shows only small areas of attenuation/atrophy in the EZ-RPE and RPE-BM (Figure 1C–D) maps compared to the nonresponder eye which demonstrates diffuse attenuation/atrophy (Figure 2C–D).

Figure 1.

Representative Risuteganib Responder OCT Foveal B-scan with anatomic layer segmentation (A) with ILM-RPE thickness map (B), EZ-RPE thickness map (C), and RPE-BM thickness map (D). The color scales on the left applies to the ILM-RPE map, and the scale on the right applies to the EZ-RPE and RPE-BM maps. The maps and B-scan demonstrate mild anatomic perturbations.

Figure 2.

Representative Risuteganib Non-Responder OCT Foveal B-scan with anatomic layer segmentation (A) with ILM-RPE thickness map (B), EZ-RPE thickness map (C), and RPE-BM thickness map (D). The color scales on the left applies to the ILM-RPE map, and the scale on the right applies to the EZ-RPE and RPE-BM maps. The maps and B-scan demonstrate significant outer retinal atrophy.

Baseline Feature and Biomarker Exploration for Treatment Response

Following initial assessment, a treatment response threshold analysis was performed, baseline EZ-RPE CST values were > 35μm in 9 of 11 eyes (82%) that responded to risuteganib compared to 3 of 12 eyes (25%) that did not respond (p<0.001). In treated eyes without GA at baseline, 10 of 16 eyes (56%) were responders to risuteganib, while only 1 of 5 eyes (20%) with GA at baseline were treatment responders (p<0.001).

Change in Compartmental OCT Metrics Over Time by Risuteganib Responder Status

No anatomical measurements showed a significant difference in the change from baseline at Week 32 between risuteganib responder eyes and non-responder eyes, except for the change in subRPE compartment volume with the responder eyes had a decline and the non-responder eyes had an increase in subRPE compartment volume (−0.049 vs 0.037 mm3; P=0.034).

Duke Reading Center Imaging Biomarker Comparison

Anatomical Measurements at Baseline by Risuteganib Responder Status

Eyes that responded to risuteganib had significantly greater mean baseline outer retinal CST (ONL-RPE) compared with eyes that did not respond to risuteganib (139.600 vs 113.917 μm; P=0.001). Responder eyes also had significantly greater baseline mean CST photoreceptor layer compared with non-responder eyes (ELM-RPE; 49.300 vs 45.083 μm; P=0.015). In addition, the en face EZ defect area of responder eyes was significantly smaller at baseline than that of non-responders (0.111 vs 0.308 mm2; P=0.012, Table 3). There were no other baseline retinal compartment measurement differences between risuteganib responder and nonresponder eyes.

Table 3:

Quantitative Anatomical Measurements at Baseline for Risuteganib Nonresponder Eyes Versus Responder Eyes

Measurement Layer, Sector	Risuteganib Nonresponder n=12	Risuteganib Responder n=10	T-test P-value
Mean thickness, μm
Inner retina, foveal center	27.833	42.500	0.305
Inner retina, central subfield	89.000	99.400	0.323
Outer retina, foveal center	124.417	143.200	0.210
Outer retina, central subfield	113.917	139.600	0.001
Photoreceptor, foveal center	46.833	48.500	0.784
Photoreceptor, central subfield	45.083	49.300	0.015
RPEDC, foveal center	47.667	58.900	0.540
RPEDC, central subfield	46.500	54.800	0.611
Total volume, mm3
Inner retina, central subfield	0.070	0.078	0.319
Outer retina, central subfield	0.090	0.110	0.001
Photoreceptor, central subfield	0.035	0.039	0.011
RPEDC, central subfield	0.037	0.043	0.600
EZ defect area, mm2	0.308	0.111	0.012

Abbreviations: EZ, ellipsoid zone; RPEDC, retinal pigment epithelium-drusen complex.

Baseline Feature and Biomarker Exploration for Treatment Response

There was macular GA in 6 of 25 (24%) and GA within the central subfield in 7 of 25 (27%) risuteganib-treated eyes. There was a 56% responder rate in risuteganib-treated eyes without any GA at baseline (n=18) compared with a 29% responder rate among risuteganib eyes with any GA at baseline (n=7) (p<0.001). There was a 58% responder rate (≥8-letter improvement threshold) in risuteganib-treated eyes without GA in the central subfield at baseline (n=19) compared with a 17% responder rate among risuteganib eyes with GA in the central subfield at baseline (n=6) (p<0.001).

Change in OCT Measurements Over Time by Risuteganib Responder Status

Quantitative analysis was also performed to measure changes in anatomical measurements over time. From baseline to Week 32, the central subfield of the inner retina in the risuteganib responder eyes had significantly larger increases in thickness (difference of 7.5 μm; P=0.042) and in volume (difference of 0.006 mm3; P=0.033) from baseline compared with risuteganib non-responder eyes. Of note, the inner retinal compartment was not assessed in the Cole analysis. There were no other significant differences in OCT measurements over time between responder and non-responder eyes.

DISCUSSION

This study highlights higher-order imaging features associated with therapeutic response to risuteganib in intermediate dry AMD without subfoveal atrophy with vison loss. Across two independent reading centers, outer retinal integrity represented by increased total and outer retinal thickness parameters and decreased quantitative ellipsoid zone disruptions correlated with responsiveness to risuteganib. Anatomically, there were no differences between the treatment and sham group across any parameter segmented which confirmed that OCT features were randomly distributed in treatment and control groups. These findings are of particular importance to the retinal disease landscape for multiple reasons. First, these findings demonstrate the potential of differentiating responders from non-responders to risuteganib therapy. Additionally, the findings support the significant impact of higher order imaging biomarkers on treatment selection and clinical trial enrichment, a personalized medicine approach.

One challenge in trials investigating such therapeutic agents relates to appropriate study endpoint for clinical efficacy and relatedly subject selection. An endpoint must have clinical relevance and be determinable through an appropriately powered study; for dry AMD prior literature has emphasized that at least a 5 letter gain or more was considered significant in eyes with vision better than 20/100 per Snellen, though for future studies and for approval a comparison of significant gain (defined by 15 letters or more) may be required for the phase III trial for clinical efficacy.^27,28 Dry AMD which has a significant spectrum of disease before developing to advanced stages involving either central geographic atrophy and/or neovascularization thus represents a challenge in trial design as subject phenotype may vary substantively within the intermediate AMD classification itself.

The current study used SD-OCT data from the risuteganib phase IIa trial to investigate whether targeted imaging phenotype extraction could differentiate likely anatomic responders to treatment. SD-OCT imaging biomarkers such as hyperreflective foci or subretinal drusenoid deposits have already previously been associated with AMD disease progression.^29,30 In the reported data, EZ integrity and sustained outer and total retina thickness were associated with increased responsiveness to risuteganib from data analyzed by two independent reading centers. The phenotype of non-responders to risuteganib is characterized by increased outer retinal thinning and loss of EZ integrity. This result suggests that clinical efficacy of risuteganib is more likely in intermediate dry AMD eyes with visual acuity loss that is not associated with more significant anatomic changes. What these associations translate to mechanistically is unclear, but it is possible that dry AMD eyes with relatively limited outer retinal atrophy have yet to reach a threshold of mitochondrial aging or OS-induced damage that would prevent risuteganib from having visually significant benefit. In fact, there may be a component of required “retinal reserve” or “photoreceptor reserve” to respond to the therapy in eyes that have not crossed a critical “atrophic” threshold. Measuring EZ integrity may be a key surrogate for this photoreceptor reserve. Additionally of note, the longitudinal changes in drusen volume were significantly different between responders and non-responder eyes. Specifically, the responder group demonstrated a reduction in the sub-RPE compartment volume whereas the non-responders demonstrated increased volume. The overall association of this finding with treatment response merits further exploration in a larger clinical trial. Given the small sample size, the significance of this finding is unclear.

Given the wide array of features on OCT that have been associated with prognosis, it is clear that such imaging represents a wealth of data to assist in clinical decision making and additionally in the investigation into novel drugs. Indeed, several imaging biomarkers identified through OCT have been associated with AMD progression or subtypes including hyperreflective foci, drusen, reticular pseudodrusen, and subretinal fluid.^30–32 Quantification of ellipsoid zone integrity using an OCT segmentation platform across eyes with dry AMD revealed that reduced integrity correlated with progression to subfoveal geographic atrophy.¹⁹ In addition to qualitative EZ integrity assessment or simple linear single B-scan assessment, prior studies have also evaluated EZ integrity with other methodologies such as EZ normalized reflectivity in en face OCT slabs at the photoreceptor layer correlating with EZ integrity and segmentation platforms combined with principle components and an Adaboost algorithm to classify EZ damage using voxels.^33,34 The potential prognostic importance that ellipsoid zone health and other such OCT biomarkers may portend in AMD progression and classification is notable and otherwise undetectable without in-depth OCT feature evaluation. Based on current literature review, such OCT segmentation platforms or imaging phenotype characterization have not been used previously in clinical trials to evaluate eyes with dry AMD. The results from the current analysis suggests that there is an important opportunity to integrate such analyses utilizing baseline imaging biomarkers into trial design to increase likelihood of discovering clinically impactful therapeutics which can be tailored to particular patient populations. Threshold analysis of imaging biomarkers may serve important role to enrich trial participant selection based on distinct anatomic threshold variance between responders and non-responders to a given therapeutic, such as risuteganib in this study.

The opportunity to use automated machine learning enhanced platforms rather than laborious manual methods to quantify imaging inclusion and exclusion criteria warrants particular note. In clinical or research practice, quantitative segmentation of higher-order imaging features would be time consuming, difficult and logistically burdensome if completed completely manually. Advances in imaging analysis methodologies have shifted this paradigm and multiple groups have highlighted the diverse applications of such tools utilizing machine-learning based platforms.^{15,16,21,35–37} The present study presents a validation of two independent masked reading centers highlighting the feasibility of incorporating such technologies into trial design.

This study has important limitations. Notably this data is sourced from a phase IIa trial which had a modest sample size given its inherently preliminary nature. As such, the insights gleaned from this data must be considered in this context. This was a post-hoc analysis and is limited by its exploratory nature. As a larger clinical trial is in the planning stages such analyses shall be forthcoming in future larger studies. These larger analyses would permit more nuanced statistical methodologies including regression testing. Additionally, the biologic underpinnings of the differences in response to risuteganib based on anatomic findings needs to be further explored in larger analyses. It is possible that EZ and outer retinal preservation also provides greater potential placebo response. The limited sham response makes this less likely, but given the small sample size and shorter sham follow-up period requires that this possibility be considered.

Overall, this report has important implications for imaging biomarkers in dry AMD and the potential utility of advanced analysis technologies in feature characterization for clinical trial enrichment. This analysis identified likely responders to risuteganib as those eyes with less outer retinal attenuation and atrophy. Based on threshold analyses, these feature extraction techniques demonstrated the potential for utilizing specific metrics for identification of eyes at baseline with the greatest potential for response to risuteganib. This supports the potential to apply higher order imaging biomarkers in subsequent trials to investigate responses of future drug candidates that could engender personalized medical treatments in diseases with significant anatomic heterogeneity. Such an application could substantially alter the ophthalmic treatment investigation paradigm, possibly enabling more efficient clinical trials and enhanced risk stratification.

Abbreviations:

(OCT)

Optical coherence tomography

(BCVA)

Best-corrected visual acuity

(AMD)

Age-related macular degeneration

(NNVAMD)

Non-neovascular AMD

(GA)

Geographic atrophy

(ILM)

Internal limiting membrane

(ONL)

Outer nuclear layer

(EZ)

Ellipsoid zone

(ELM)

External limiting membrane

(RPE)

Retinal pigment epithelium

(BM)

Bruch’s membrane

(RPEDC)

RPE drusen complex