You just need the data in the format:

1    160283    471362    DUP
1    1385015   1387562   DEL

[source: https://www.ensembl.org/info/docs/tools/vep/vep_formats.html]

So, you need to set rules about copy number gain and loss based on the segment_mean column in your data. I have seen values of -0.2 / +0.2 and -0.3 / +0.3 used, in the past.

If, instead, you are only interested in which genes overlap your regions right away, then take a look at my answer here: How to extract the list of genes from TCGA CNV data

Further down on that page ( C: How to extract the list of genes from TCGA CNV data ), I then actually go over how you can merge your copy number aberrations into recurrent somatic coy number alterations (recSCNA) and show how you can eventually end up with regions as AMP (DUP) or DEL, which is suitable then for VEP.

Kevin