Diabetic macular edema (DME) remains a leading cause of vision loss worldwide. DME is commonly treated with intravitreal injections of vascular endothelial growth factor (VEGF)-neutralizing antibodies. VEGF inhibitors (anti-VEGFs) are effective, but not all patients fully respond to them. Given the potential side effects, inconvenience, and high cost of anti-VEGFs, identifying who may not respond appropriately to them and why is essential. Herein we determine first the response to anti-VEGFs, using spectral-domain optical coherence tomography scans obtained from a cohort of patients with DME throughout the 1st year of treatment. We found that fluid fully cleared at some time during the 1st year in 28% of eyes ("full responders"); fluid cleared only partly in 66% ("partial responders"); and fluid remained unchanged in 6% ("nonresponders"). To understand this differential response, we generated induced pluripotent stem cells (iPSCs) from full responders and nonresponders, from subjects with diabetes but no DME, and from age-matched volunteers without diabetes. We differentiated these iPSCs into endothelial cells (iPSC-ECs). Monolayers of iPSC-ECs derived from patients with diabetes showed a marked and prolonged increase in permeability upon exposure to VEGF; the response was significantly exaggerated in iPSC-ECs from nonresponders. Moreover, phosphorylation of key cellular proteins in response to VEGF, including VEGFR2, and gene expression profiles, such as that of neuronal pentraxin 2, differed between full responders and nonresponders. In this study, iPSCs were used in order to predict patients' responses to anti-VEGFs and to identify key mechanisms that underpin the differential outcomes observed in the clinic. This approach identified NPTX2 as playing a significant role in patient-linked responses and as having potential as a new therapeutic target for DME.