G-CSF rescued the vascular change in the OIR model. (A) Time scheme of the experiments. G-CSF (50 μg/kg) was administered intraperitoneally daily between P6 and P10. (B-F) Retinal whole mounts stained with collagen IV antibody to detect the area with retinal vasculature. (B-C) Control P7 and P12, respectively. Note the nonvascular area (white) in the P12 retina that was not observed in the P7 retina. Vascular obliteration occurred between P7 and P12 in this model. (D) A G-CSF–treated P12 retina. The area of vascular obliteration (white) was decreased by G-CSF. (E-F) At P17, neovascular tufts (yellow) and vascular attenuation (white) were marked in the control OIR retina (E), and G-CSF reduced both changes (F). (G) G-CSF injection reduced the obliteration area at P12 compared with the control (n = 13 each group). **P < .01. (H-I) G-CSF reduced both the vascular obliterated area and the neovascular tuft at P17 compared with the controls (n = 13 in each group). *P < .05. **P < .01. (J-L) Intravitreal injection of G-CSF (P6) also reduced the vascular attenuation (white) in the OIR model. (J) Control treated retina at P12. (K) G-CSF-treated retina at P12. Intravitreal G-CSF injection reduced the obliteration area significantly to that of control (n = 5 each group). *P < .05. (M-O) Vehicle-treated (M) and G-CSF–treated (N) OIR retina, stained with CD45 antibody, and counted the CD45⁺ cells at 4 points of 300 μm² in the avascular area at P12. No remarkable difference in the number of CD45⁺ cells. (O) n = 4 each group. P > .05.