Depending on your question, the significance of the estimated true effect may be less important than the estimated residual heterogeneity (tau^2). You should run the random effects model for your analysis and see if the test of residual heterogeneity (H0: tau^2 = 0) is significant (reject H0). If so, you'd conclude there is significant heterogeneity among the true effects and thus you'd be interested in conducting a meta-regression to explain that heterogeneity.