Orientation bias in variants called by mutect2

Question

[bt_cepo]

Hi! I am performing variant calling analysis with mutect2 in tumor-only mode for a ctDNA sample. Checking the output I see variants in my gene of interest, but they do not have the PASS label in the FILTER column but "orientation", stating that they have orientation bias in the reads. I want to ask if these variants are still reliable even though they have this bias, or IF I have to disregard them completely.

The sample I am analyzing presents 100% for sure variants in the gene of interest (I am doing benchmarking) so if these variants are not real I am not sure what kind of optimization for ctDNA I could include in the pipeline.

Thank you!! :)