Objective To analyze the consistency of clinical imaging and clinicopathological finds of retinoblastoma (RB) optic nerve invasion. Methods A retrospective case study. Fifteen children with 15 eyes who were diagnosed with RB and underwent enucleation at the Eye Center of Beijing Tongren Hospital from November 2017 to January 2022 were included in the study. Among them, there were 9 males with 9 eyes and 6 females with 6 eyes. The mean age was 1.75±1.61 years. All affected eyes were designated International Classification of Retinoblastoma group E. There were 7 cases with secondary neovascularization glaucoma, 2 cases with closed funnel-shaped detachment of the retina and tumor touching the posterior capsule of the crystal, and 6 cases with tumor touching the back surface of the crystal and posterior chamber of the 15 children. All children underwent CT or magnetic resonance imaging (MRI). Among them, CT examination was performed in 4 cases, MRI examination in 4 cases, and MRI and CT examination in 7 cases. All the children underwent eyeball enucleation, paraffin sections were taken from the eyeball, and sagittal section of the eyeball with optic nerve tissue was taken for pathological examination. Imaging diagnosis was based on optic nerve thickening and/or enhancement. The pathological diagnosis was based on the growth of RB tumor cells across the post-laminar of optic nerve. The pathological diagnosis was based on the growth of RB tumor cells across the sieve plate. The area under the curve (AUC) of receiver operating characteristic curve was used to evaluate the diagnostic ability of imaging examination. The sensitivity, PPV and 95% confidence interval (CI) of imaging examination were obtained based on the confusion matrix. With pathology as the gold standard, intragroup correlation coefficient (ICC) was used to test the consistency of imaging and pathology. Results All the 15 cases were diagnosed with RB by pathological examination. Of the 15 cases, 7 cases were diagnosed with RB optic nerve invasion by imaging and 8 cases without nerve invasion; 12 cases of RB optic nerve invasion were diagnosed by pathology after operation. Three cases without nerve invasion. Among them, 4 cases had identical imaging and pathological findings. The sensitivity and PPV of MRI and CT were 0.33 (95%CI 0.11-0.64) and 0.57 (95%CI 0.20-0.88), respectively. The value of AUC (the area under the curve) in the ROC curve of MRI and CT were 0.51 (95%CI 0.24-0.77) and 0.52 (95%CI 0.25-0.78), respectively. The AUC values of both were 0.5-0.7, indicating low diagnostic accuracy. ICC test evaluated the consistency of MRI and CT with pathological examination, 0.61 (95%CI 0.97-0.87) and 0.63 (95%CI 0.12-0.88) for MRI and CT, respectively. Therefore, the consistency of MRI, CT and pathology was moderate. Conclusion Compared with the pathological findings, the sensitivity and accuracy of MRI and CT in the diagnosis of RB optic nerve post-laminar invasion are lower, and the consistency between MRI and CT with pathology is only moderate.

Citation： Zhao Hongshu, Wei Wenbin, Zhang Xu. Analysis of the concordance between imaging and histopathologic finds in patients with retinoblastoma optic nerve invasion. Chinese Journal of Ocular Fundus Diseases, 2023, 39(1): 5-10. doi: 10.3760/cma.j.cn511434-20220408-00203 Copy