# OCTA‐Derived Retinal Biomarkers and Infarct Topography Improve Etiologic Classification of Recent Single Subcortical Infarction: A Nomogram Model

## TL;DR

A new model combining retinal scans and brain imaging improves diagnosis of stroke causes in patients with recent subcortical infarction.

## Contribution

A novel nomogram integrating OCTA retinal biomarkers and infarct topography improves etiological classification of RSSI.

## Key findings

- The nomogram achieved a C-index of 0.84, outperforming conventional imaging (C-index 0.68).
- Three key variables—superficial vascular complex density, lesion slices, and proximal lesion location—were critical for differentiation.
- The model may support etiology-driven clinical decisions for RSSI patients.

## Abstract

Recent single subcortical infarction (RSSI) in lenticulostriate artery territories exhibits etiological heterogeneity. Misclassification risks persist due to overlapping neuroimaging features between cerebral small‐vessel disease‐related lacunar infarction (CSVD‐related LI) and branch atheromatous disease (BAD). We developed a nomogram that integrates retinal optical coherence tomography angiography (OCTA) metrics with infarct topography to improve etiological classification.

Patients with RSSI were prospectively enrolled between December 2021 and December 2023. LASSO regression identified predictors for a logistic regression–based nomogram. Performance was evaluated via concordance index (C‐index), calibration curves, and decision‐curve analysis.

A total of 127 RSSI patients (86 CSVD‐related LI, 41 BAD) were included. Three variables—superficial vascular complex density, number of lesion slices, and proximal lesion location—were retained in the final model. The nomogram achieved a C‐index of 0.84 (95% CI, 0.80–0.89) versus 0.68 for conventional imaging, with superior net benefit across clinical thresholds (AUC 0.84 vs. 0.68, p < 0.001).

The novel nomogram combining OCTA‐derived retinal biomarkers with infarct topography improves differentiation of BAD from CSVD‐related LI in RSSI patients and may facilitate etiology‐driven clinical decision‐making. External validation is needed for clinical implementation.

A nomogram integrating reduced superficial vascular complex density on OCTA with proximal and multi‐slice infarct topography improves differentiation of branch atheromatous disease (BAD) from CSVD‐related lacunar infarction (CSVD‐related LI) in recent single subcortical infarction (RSSI).This retina‐enhanced model outperforms conventional imaging alone and may facilitate etiology–driven clinical decision‐making.

A nomogram integrating reduced superficial vascular complex density on OCTA with proximal and multi‐slice infarct topography improves differentiation of branch atheromatous disease (BAD) from CSVD‐related lacunar infarction (CSVD‐related LI) in recent single subcortical infarction (RSSI).

This retina‐enhanced model outperforms conventional imaging alone and may facilitate etiology–driven clinical decision‐making.