To perform an unsupervised machine learning clustering of patients with punctate inner choroidopathy (PIC) and provide new insights into the significance of pachychoroid disease features in PIC eyes.
Retrospective multicenter study, including 102 eyes from 82 patients diagnosed with PIC. Demographics, clinical data, and multimodal imaging (MMI), including fundus photography, optical coherence tomography (OCT), and indocyanine green angiography (ICGA), were collected. Clusters of eyes were identified, and multilevel logistic regression analysis was performed to compare between-group differences.
Using 17 clinical features, two distinct PIC patient clusters were identified. Cluster 1 was characterized by older age, high myopia, myopic maculopathy features, thin choroids, multiple lesions, and a higher likelihood of developing patchy chorioretinal atrophy. Cluster 2 consisted of younger age, emmetropia or low myopia, thick choroids, choroidal vascular hyperpermeability on late-phase ICGA, and a high prevalence of focal choroidal excavation. These features exhibited significant differences (p<0.05) between the two clusters.
While PIC typically affects young myopic females with thin choroids, a subset of PIC patients exhibits features associated with pachychoroid disease. Considering the potential influence of choroidal venous insufficiency on PIC manifestations and secondary complications, we propose the term “punctate inner pachychoroidopathy” to characterize this distinct subtype of PIC.