Batch Effects in Tumor-Only NGS Panel Sequencing and Implications for CNV Detection

Next generation sequencing (NGS) is routinely performed in clinical practice to detect various types of mutations for targeted therapy, diagnosis, and prognosis. Actionable alterations detected by NGS include not only non-synonymous mutations that lead to functional or structural changes of proteins but also copy number variants (CNV) that affect gene dosage, such as gene copy gains, amplifications or deletions. Among tumor-only CNV detection methods, the use of a Panel of Normals (PoN) for relative comparison has become a common practice, largely due to the lack of matched normal samples. It was therefore hypothesized, once established, a PoN and CNV caller may not fully compensate for all experimental variations - such as differences in probe efficiency across reagent lots. To investigate this, 12,104 clinical sequencing datasets from 1,454 sequencing batches were analyzed over a four-year period. This analysis revealed batch-associated fluctuation patterns in gene-level fold changes that could potentially lead to misinterpretation, such as the incorrect classification of gene copy deletions or gains. In this study, a strategy is presented that calculates the median and median absolute deviation of gene-level fold changes across all samples within each sequencing batch and incorporates these measures into the result interpretation. By providing batch-level reference metrics, putative batch-driven artifacts can be identified, reducing false-positive CNV calls and supporting more reliable interpretation in comprehensive genomic profiling.