CLEAR report 1: a scoping review and meta-analysis for definitions, imaging metrics, and functional correlates of...

Advanced age-related macular degeneration (AMD) is a leading cause of irreversible vision loss worldwide, with photoreceptor degeneration representing the final common pathway of functional impairment. Optical coherence tomography (OCT) enables noninvasive, layer-resolved quantification of photoreceptor integrity using biomarkers of the ellipsoid zone (EZ), the outer nuclear layer (ONL), and the external limiting membrane (ELM). However, substantial heterogeneity in definitions, measurement protocols, and reporting practices limits cross-study comparability and the adoption of clinical trials. This scoping review with nested meta-analyses maps how axial photoreceptor biomarkers are operationalized, segmented, validated, and related to functional and multimodal endpoints in AMD.

MEDLINE, Embase, and Scopus were searched from January 2015 to December 2025 following PRISMA-ScR guidelines. Eligible studies reported OCT-based photoreceptor biomarkers in AMD populations. Data extraction captured boundary definitions, imaging platforms, segmentation approaches, analytic domains, ROI strategies, phenotypic context, reliability statistics, structure-function correlations, and OCT-FAF agreement.

Ninety-four studies met inclusion criteria, spanning spectral-domain and swept-source OCT platforms and diverse AMD phenotypes. Marked variability was observed in boundary definitions (e.g., EZ-RPE vs. EZ-Bruch's membrane), measurement strategies (thickness, area, volume, reflectivity), and spatial sampling (central 1 mm, ETDRS subfields, lesion-centric masks). When segmentation boundaries and analytic conventions were explicitly defined and consistently applied, reliability for EZ and ONL metrics was high (ICC > 0.90), with automated and deep-learning methods performing comparably to expert manual grading. Structure-function analyses demonstrated moderate-to-strong correlations between photoreceptor loss and microperimetry sensitivity (r = 0.50-0.80) and best-corrected visual acuity (r = 0.40-0.70), particularly in advanced disease stages. OCT-FAF agreement for geographic atrophy detection was similarly high (κ > 0.80), although early atrophic features showed greater definitional variability. Phenotypic stratification and topographic reporting were inconsistently documented.

OCT-derived photoreceptor biomarkers show high analytical reliability and clinically meaningful structure-function associations when segmentation boundaries, analytic domains, and reporting conventions are clearly specified. However, the review identifies substantial construct heterogeneity that complicates cross-study comparability and endpoint interpretation. Adoption of consensus boundary definitions, minimum reporting standards, and harmonized validation frameworks is necessary to ensure reproducibility, facilitate regulatory evaluation, and enable integration of photoreceptor metrics into AMD prevention and early-intervention trials. These findings provide the conceptual foundation for the CLEAR initiative (Consensus Layer Evaluation for AI Algorithm Reporting).