Predictive Factors for Vision Recovery after Optic Nerve Decompression for Chronic Compressive Neuropathy: Systematic Review and Meta-Analysis

Background: Optic nerve decompression for chronic compressive neuropathy results in variable success of vision improvement after surgery. Factors predicting visual improvement are variable in the published literature due to small samples and single-center differences. We sought to determine the effect of various factors through meta-analysis of available literature.

Methods: A MEDLINE search was performed for years 1990-2010 for articles with preoperative and postoperative vision data in the following categories: tumor (meningioma, craniopharyngioma, pituitary, skull base), aneurysm, mucocele, and fibrous dysplasia. Case reports; non-English articles; and articles addressing other pathologies such as trauma, pseudotumor, glaucoma, and central retinal vein occlusion were excluded. Articles were then categorized by data quality. Three factors-size of tumor, approach (transsphenoidal vs. craniotomy), and duration of preoperative vision loss—were chosen for initial analysis, and rate of improvement was collected for each study. Odds ratios and 95% confidence intervals for each study were calculated using exact methods as implemented in SAS v9.2 (proc logistic). Mantel-Haenszel (M-H) odds ratios were used to summarize overall effects, and Breslow-Day tests were used to assess homogeneity of odds ratios. Random effects logistic regression was used to assess associations with outcomes when there was significant variation among odds ratios.

Results: After 273 unique citations were screened, a review of references revealed an additional 71 citations to screen. Forty articles met inclusion criteria for this initial analysis, and 15 factors were identified that were analyzed in more than one article. Tumor size had 12 studies (n = 467), with an overall M-H OR of 1.82 (95% CI 1.22-2.73) for recovery with smaller tumors compared with larger tumors. Approach (6 studies, n = 465) did show some study heterogeneity (chi² 14.9, p = .01), and a random effects OR of 0.37 (95% CI 0.20-0.68) favored transsphenoidal surgery for vision recovery. Duration of symptoms (11 studies, n = 485) strongly favored shorter duration with M-H OR 3.75 (95% CI 2.45–5.74) for recovery.

Conclusions: Meta-analysis can be a robust method to detect effects that may better predict outcome when there is disagreement in published literature. Initial analysis shows that small size, shorter duration of symptoms, and transsphenoidal approach have higher odds of improvement of vision.