ObjectivesTo observe the changes in the thickness of the retinal nerve fiber layer (RNFL), ganglion cell layer (GCL), and inner plexiform layer (IPL) after trabeculectomy for primary open-angle glaucoma (POAG), and preliminarily analyze their relationship with visual function. MethodsA retrospective clinical study. From January 2022 to March 2025, 103 patients (103 eyes) with POAG who underwent trabeculectomy at the Department of Ophthalmology of Anqing Municipal Hospital were enrolled in the study. All affected eyes underwent best corrected visual acuity (BCVA), visual field, and optical coherence tomography (OCT) examinations. Postoperative BCVA decline of ≥1 line, or new defects in the visual field, deepening or expansion of existing scotomas, was defined as visual function decline. Based on whether visual function declined one month after surgery, the affected eyes were divided into the non-decline group and the decline group, with 58 and 45 eyes, respectively. OCT was used to measure the thickness of the macular region, including the superior, inferior, temporal, and nasal RNFL, GCL, and IPL. According to the diabetic retinopathy treatment research group's classification, the retina within a 6 mm radius of the macular fovea was divided into three concentric circles centered on the macular fovea, namely the central zone with a diameter of 1 mm, the inner ring zone ranging from 1 to 3 mm, and the outer ring zone ranging from 3 to 6 mm, for a total of nine zones. The four zones of the inner and outer rings were designated as superior (S1, S2), inferior (I1, I2), nasal (N1, N2), and temporal (T1, T2), respectively. The changes in GCL, IPL, and RNFL thickness in different macular regions of the two groups of affected eyes were compared and observed. Logistic regression and restricted cubic spline methods were used to analyze the relationship between GCL, IPL, RNFL thickness and postoperative visual function recovery; the area under the receiver operating characteristic curve (ROC curve) (AUC) and decision curve (DCA) were used to evaluate the predictive performance and clinical benefit of the multi-parameter retinal thickness combined detection model for postoperative visual function. ResultsCompared with before surgery, the average and regional GCL and RNFL thicknesses in the macular area significantly increased after surgery, with statistically significant differences (P＜0.05). After adjusting for age, intraocular pressure, hyphema, anterior chamber inflammation, and visual field stage, the average [odds ratio (OR)=3.413, 95% confidence interval (CI) 1.948?5.978] and T1 (OR=4.949, 95%CI 2.373?10.321), I1 (OR=5.105, 95%CI 2.571?10.137), I2 (OR=2.672, 95%CI 1.332?5.358) regional GCL thicknesses, as well as I1 (OR=3.784, 95%CI 2.033?7.043) and I2 (OR=3.152, 95%CI 1.417?7.013) regional IPL thicknesses, and T1 (OR=3.101, 95%CI 1.719?5.594) and T2 (OR=4.110, 95%CI 2.165?7.802) regional RNFL thicknesses were significantly correlated with postoperative visual function (P＜0.05). The association strength between average GCL thickness and postoperative visual function exhibited a nonlinear dose-response relationship (P＜0.05). The results of ROC curve analysis showed that AUC for predicting postoperative visual function recovery in POAG eyes was greater than 0.7 for the average macular area and T1, I1 regional GCL thicknesses, I1 regional IPL thickness, and T1, T2 regional RNFL thicknesses; the AUC for combined prediction using GCL, IPL, and RNFL thicknesses was 0.926, with sensitivity and specificity of 0.942 and 0.916, respectively. The results of DCA analysis indicate that the combined prediction of GCL, IPL, and RNFL thickness has the largest range of net benefit thresholds, suggesting that the combined net benefit of the three is the highest. ConclusionsThe recovery of visual function after trabeculectomy in POAG-affected eyes is correlated with the thicknesses of GCL, RNFL, and IPL in the macular region, especially in the temporal and inferior areas. The combined prediction of postoperative visual function recovery using multiple retinal thickness parameters is more effective.