Survivin overexpression is potentially associated with pituitary adenoma invasiveness

BACKGROUND AND OBJECTIVE: Survivin is an inhibitor of apoptosis. Its role in guiding the treatment of neoplasms, making diagnosis and predicting prognosis has been reported. However, there is little information on the implications and uses of survivin in predicting pituitary adenoma (PA) invasiveness. Existing information is unclear and controversial. We thus conducted this meta-analysis to explore whether the surviving expression levels in invasive PAs (IPA) and regular PAs are different or not. We considered both non-secreting and secreting tumors together. METHODS: A global search strategy was systematically applied among five databases including Cochrane Library, Embase, PubMed, Web of Science, and Chinese National Knowledge Infrastructure (CNKI) up to June 18th, 2017. With a specially designed form including PAs’ invasive features, etc., data was collected. The included studies should present the data representing the surviving levels in IPA groups and regular PA groups, respectively. Differences were expressed as standard mean differences (SMDs) or odds ratios (ORs) with 95% confidence interval (CI). To estimate the heterogeneities, I(2) test, Cochran's Q-test and Galbr figure were all conducted. A sensitivity-analysis and potential-publication bias were also performed. RESULTS: In the present meta-analysis, 9 studies containing 489 patients were included. Seven studies with dichotomous-data showed that survivin over-expression in PA tissue was closely associated with a high invasive tendency (OR 6.226, 95% CI 3.970, 9.765; P<0.001), but 2 continuous-data studies revealed that there was no significant association (SMD −5.043, 95% CI-10.965, 0.878; p=0.095). A sensitivity-analysis suggested a statistically stable result. We did not find publication bias. CONCLUSION: We suggest that survivin overexpression is potentially associated with PA invasiveness. More research based on medical big data is needed to confirm this finding.