WGCNA is not a technique for finding significant genes. Instead it is a clustering alogrithmn that clusters genes with correlated gene expression across samples together. Modules of genes are created using a dynamic tree cutting approach which doesn't have a hard threshold. That is, there is no statistical significance (in the p-value sense) attached to a gene being associated with a module. Are you looking to extract the genes that belong to each module?

If you want to find significantly different genes between two conditions I recommend using an algorithm designed for that purpose. The most common is limma.