Precision Mapping of Retinal Disease: Identification of Differential Progression Trajectories via Imaging Phenomics

ABSTRACT We propose a new framework for the analysis of large clinical phenotypic datasets through the repurposing of single‐cell RNA‐sequencing analysis techniques. By utilizing similarities between cellular development and clinical heterogeneity, we show that techniques such as clustering, pseudo‐time, and trajectory analysis can reveal different disease progression trajectories. Our approach permits the extraction of long‐term progression trajectories from large cross‐sectional datasets while overcoming the hurdle of a lack of longitudinal data plaguing many large clinical phenotypic studies. We present this framework on the retinal degenerative disease macular telangiectasia type 2 (MacTel) and derived previously unknown differential disease progression trajectories from 87 clinically graded phenotypic variables measured in 7888 eyes (3971 patients). One progression trajectory was characterized by vascular alterations, the other by severe neurodegeneration and vision loss. Older and diabetic patients were significantly more likely to progress toward the neurodegenerative route. We demonstrate how pseudo‐time calculations reveal a holistic fine‐scale data‐driven progression score useful for analytical investigations, which complements previously established clinical severity scores. Lastly, integration of genetic data on 71% of patients revealed that genetic background affects phenotype heterogeneity, differential progression trajectory, but not progression rate.