Classifications Proposed for Pachychoroid Disease

This study highlights a need for the classification of pachychoroid disease. Currently, the spectrum of this disease covers various chorioretinal disorders, such as central serous chorioretinopathy, pachychoroid pigment epitheliopathy, pachychoroid neovasculopathy, peripapillary pachychoroid syndrome and polypoidal choroidal vasculopathy. However, not much is understood about this disease spectrum, as these disorders have not been officially deemed as distinct diseases or forms of progression. Therefore, categorizing disease progression into different clusters can help provide a framework for future research. Photo: Jessica Haynes, OD, and Mohammad Rafieetary, OD. Click image to enlarge. Thanks to the latest advancements in imaging technology, clinicians are able to examine ocular diseases from a new perspective. This opens the gateway for researchers to identify new phenotypes of conditions that have very little understanding in the field.In a recent study published in Eye, researchers shared their method for classifying different types of pachychoroid disease (PD). Using patient data, OCT and fundus autofluorescence imaging, they were able to define four unique classifications of the disease. Machine learning models were used to achieve the study’s primary and secondary objectives. Additionally, the impact of photodynamic therapy on disease progression was studied.After screening 1,208 electronic health records from patients with a history of PD, researchers identified 973 eyes (663 patients) with acceptable coding, imaging data and five-year follow-up history. Using baseline information, the cluster model outlined four specific groups of PD:Cluster 1: Active Mild Phenotype (~20/29 Snellen)Cluster 2: Mild or Resolving Phenotype (~20/22 Snellen)Cluster 3: Mixed Chronic Phenotype (~20/45 Snellen)Cluster 4: Severe Chronic Phenotype (~20/91 Snellen)Cluster 1 (233 eyes) included patients who were younger and experiencing active fluid and mild structural alterations. Cluster 2 (317 eyes) included patients with mild or resolving PD who maintained an optimal visual acuity (VA). Cluster 3 (336 eyes) included patients with chronic fluid accumulation, significant structural damage and a somewhat poor VA. Lastly, Cluster 4 (87 eyes) included patients with severe cases and worsening VA.The researchers’ modeling showed that over a five-year follow-up period, PD progression is imminent. At baseline, PD’s prevalence was lowest for Cluster 4 (12%). After follow-up, disease progressed, and Cluster 4 grew to 27.3%. The category that was most affected was Cluster 1. This group began at 22.7%, and after follow-up, 53% of eyes progressed to Cluster 2 and 37% progressed to Cluster 3, reducing Cluster 1 to 3.7%. Then, researchers added photodynamic therapy into the model’s equation. However, no significant alterations towards disease progression were observed for patients administered this treatment.When observing VA changes, researchers noticed that the impact on vision differed by each cluster. When compared to Cluster 1, Cluster 4 had the worst baseline VA and continued to significantly worsen over time. This result was unlike those calculated by the linear-mixed effects model for Clusters 2 and 3. These categories did not present with a severe visual impairment at baseline, nor did they significantly worsen during follow-up.“This large, data-driven study identified four distinct pachychoroid phenotypes, demonstrating the potential of machine learning for objective disease classification,” concluded the researchers in their paper. “Despite significant structural evolution, VA remained stable, emphasizing the need for complementary functional assessments, patient-reported outcome measures, and more effective treatments that can modify the underlying disease course.”Click here for the journal source. Cicinelli MV, Bianco L, Caminada L, et al. Towards an objective classification of pachychoroid disease and its risk of progression. Eye. February 16, 2026. [Epub ahead of print]. This article was developed by the editorial staff in conjunction with experts in the field. In the process, AI may have been among the editorial tools used to meet the goals of human editors, who approved all content.