Optimization of Population Frequency Cutoffs for Filtering Common Germline Polymorphisms from Tumor-Only...

Area of Science:

• Genomics
• Cancer Research
• Bioinformatics

Background:

• Clinical tumor-only sequencing complicates germline versus somatic variant differentiation.
• Population frequency (PF) cutoffs lack consensus, impacting variant classification accuracy.

Purpose of the Study:

• To evaluate the impact of PF cutoffs on variant classification in tumor-only sequencing.
• To determine optimal, assay-specific PF cutoffs for improved germline and somatic variant identification.

Main Methods:

• Utilized five population databases and the Catalog of Somatic Mutations in Cancer.
• Analyzed The Cancer Genome Atlas (TCGA) and clinical hybridization capture test data.
• Assessed variant classification accuracy across different PF cutoffs and assay types.

Main Results:

• Widely used 1-2% PF cutoffs showed high somatic sensitivity but reduced germline sensitivity.
• Optimized PF cutoffs predicted variant origin in TCGA data with >95% accuracy.
• Optimal cutoffs varied by cancer type and assay region; PF cutoffs were not transferable between assays.

Conclusions:

• Filtering approaches for tumor-only testing must be assay-specific and carefully optimized.
• Optimized, assay-specific PF cutoffs enhance variant classification accuracy.
• This approach supports reliable tumor-only testing pending routine tumor-normal sequencing.