Arrigo Alessandro, Calamuneri Alessandro, Aragona Emanuela, Bordato Alessandro, Grazioli Moretti Alessio, Amato Alessia, Bandello Francesco, Battaglia Parodi Maurizio
Department of Ophthalmology, IRCCS Ospedale San Raffaele, University Vita-Salute, Milan, Italy.
University of Messina, Messina, Italy.
Ophthalmol Ther. 2021 Jun;10(2):289-298. doi: 10.1007/s40123-021-00336-3. Epub 2021 Feb 19.
This study aimed to assess quantitative factors associated with treatment response and macular neovascularization (MNV) onset in central serous chorioretinopathy (CSC) through an artificial intelligence-based approach.
The study was designed as an interventional, prospective case series with a planned follow-up of 36 months. We included only eyes demonstrating the first episode of CSC. All the patients underwent eplerenone or photodynamic therapy (PDT) treatment. Eyes developing MNV underwent anti-VEGF injections. We developed an artificial intelligence-based model to assess predictive quantitative structural optical coherence tomography (OCT) factors related to treatment response and onset of MNV. Main outcome measures were best-correct visual acuity (BCVA), central macular thickness (CMT), retinal thickness (RT), retinal pigment epithelium (RPE) thickness, choroidal thickness, Sattler's layer thickness (SLT), Haller's layer thickness, retinal and choroidal hyperreflective foci (HF), and MNV.
We included 96 naïve CSC eyes (96 patients). Baseline BCVA was 0.18 ± 0.25 logMAR, which increased to 0.16 ± 0.27 logMAR after 3 years (p > 0.05). Baseline CMT was 337 ± 126 µm, which improved to 229 ± 40 µm after 3 years (p < 0.01). We observed good response to eplerenone in 40/78 (51%) eyes, whereas 38/78 (49%) eyes underwent PDT. The artificial intelligence model showed choroidal HF and age as determining factors of good response to eplerenone or PDT. RPE thickness < 36 µm, RT < 300 µm, and SLT < 50 µm increased probability of 50% of having MNV.
CSC response to eplerenone or PDT is influenced by choroidal HF and patient age. RPE and SLT represent relevant factors for onset of MNV.
本研究旨在通过基于人工智能的方法评估与中心性浆液性脉络膜视网膜病变（CSC）治疗反应及黄斑新生血管（MNV）发生相关的定量因素。
本研究设计为一项干预性前瞻性病例系列研究，计划随访36个月。我们仅纳入首次发作CSC的患眼。所有患者均接受依普利酮或光动力疗法（PDT）治疗。发生MNV的患眼接受抗VEGF注射治疗。我们开发了一种基于人工智能的模型，以评估与治疗反应及MNV发生相关的预测性定量结构光学相干断层扫描（OCT）因素。主要观察指标为最佳矫正视力（BCVA）、中心黄斑厚度（CMT）、视网膜厚度（RT）、视网膜色素上皮（RPE）厚度、脉络膜厚度、萨特勒层厚度（SLT）、哈勒层厚度、视网膜和脉络膜高反射灶（HF）以及MNV。
我们纳入了96只初发CSC患眼（96例患者）。基线BCVA为0.18±0.25 logMAR，3年后增至0.16±0.27 logMAR（p>0.05）。基线CMT为337±126μm，3年后改善至229±40μm（p<0.01）。我们观察到78只眼中有40只（51%）对依普利酮反应良好，而78只眼中有38只（49%）接受了PDT治疗。人工智能模型显示脉络膜HF和年龄是对依普利酮或PDT反应良好的决定因素。RPE厚度<36μm、RT<300μm以及SLT<50μm会使发生MNV的概率增加50%。
CSC对依普利酮或PDT的反应受脉络膜HF和患者年龄的影响。RPE和SLT是MNV发生的相关因素。
