Varscan2 work flow from BAM producing too few somatic mutation calls?

Hi!

I am recently using the Varscan2 workflow from BAM to do somatic mutation calling of TCGA GRch38 BAM files. However, the output high confidence vcf files is only a few kb large. One of the patient I was looking, TCGA-AR-A1AO has around 6000 mutations called in the MuTect vcf but only have 300 mutations in my output.
I didn't change any parameters. I wonder if it's the problem of input files but I was just using the tumor-normal bam in TCGA.