MuTect v2 t_LOD filtering

Dear All,
we actually use GATK3.5 in combination with MuTect v2 to identify somatic variants. In the output, both FoxoG and t_LOD scores are listed. However, I see a predominance of C>A mutations (appr. 70%) and wonder, whether the variants are already filtered for oxidation induced changes as described by Costello et al. (PMID 23303777). If I do calculations for the C>A variants I get approximately half t_LOD values above the threshold (-10+(100/3)*FoxoG) - i.e. the "good ones" - and another half below (the putative artifacts). So, I suppose that filtering has not been done - is that right?
Thanks for your help - Stefan