Nab-Paclitaxel-Induced Cystoid Macular Edema with Minimal Fluorescein Leakage: A Case Report of Two Patients from a Tertiary Referral Center.

INTRODUCTION: The aim of the study was to report 2 cases of nab-paclitaxel (nab-PTX)-induced macular edema and describe its incidence over a 5-year period at a tertiary center. CASE PRESENTATIONS: We retrospectively reviewed patients who received nab-PTX between July 2020 and June 2025 across several departments at our institution. Ophthalmic findings, multimodal imaging results, management, and outcomes were analyzed. Among 183 patients treated with nab-PTX, 25 symptomatic patients were referred to ophthalmology, and 2 were adjudicated as definite taxane-induced macular edema (TIME; observed proportion, 2/183; 1.1%). Both patients were men with metastatic pancreatic cancer receiving gemcitabine and nab-PTX. Optical coherence tomography (OCT) demonstrated bilateral cystoid macular edema. In Case 1, fluorescein angiography (FA) showed no macular leakage or vasculitis. Central retinal thickness (CRT) improved from 479/582 µm (right/left) to 175/173 µm after discontinuation of therapy, with macular edema resolving by 11 weeks. In Case 2, FA was not performed because of the patient's poor general condition; OCT angiography showed no microaneurysms or other findings suggestive of alternative causes of macular edema. CRT improved from 659/663 µm to 207/232 µm at 1 month after discontinuation, with complete resolution by 3 months. Best-corrected visual acuity improved in both cases after nab-PTX discontinuation. CONCLUSION: Nab-PTX can cause bilateral cystoid macular edema, often with minimal or absent FA leakage, and treatment discontinuation can lead to anatomical and functional recovery. Early recognition and prompt interdisciplinary management are important to prevent potentially irreversible visual impairment. Because ophthalmologic evaluation was symptom-driven, asymptomatic, or mildly symptomatic, TIME may have been underdetected; therefore, this observed proportion should not be interpreted as a population-level incidence estimate.