Detection of Copy-Number Variations in CNS Tumours From Off-Target Reads of Hybrid-Capture Sequencing.

When doctors run genetic tests on tumor samples, they are typically hunting for mutations in a specific list of genes. To do this, they use a technique called hybrid-capture sequencing, which acts like a molecular fishing hook, pulling out only the DNA segments they care about. But this process also hauls in a lot of 'off-target' reads, genetic material from across the whole genome that gets sequenced incidentally and is usually discarded. Researchers wondered whether this overlooked data could be put to work.

The team tested whether these discarded off-target reads could be used to map copy-number variations, or CNVs, across the entire genome of brain tumor samples. CNVs are spots where chunks of DNA have been duplicated or deleted, and they are critically important for classifying and grading central nervous system tumors. Currently, a separate and specialized test called a methylation array is the standard way to get this genome-wide CNV picture. The researchers compared CNV profiles generated from the off-target reads of a small, inexpensive sequencing panel against the results from methylation arrays in 124 brain tumor samples spanning several tumor types. Across 527 chromosomal-scale changes, the two methods agreed 95% of the time. The approach correctly identified all 19 focal gene amplifications, such as those in EGFR and MYCN, genes whose overactivity can drive tumor growth. It also detected CNV patterns suggesting a cancer-driving gene fusion called BRAF in 5 out of 6 tumor samples known to carry it. Notably, in meningiomas, a common type of brain tumor, the method uncovered molecular abnormalities that warranted upgrading the severity classification in 16% of cases that had appeared lower-grade under the microscope alone. The study does note that the method missed some instances of a specific type of deletion in a gene called CDKN2A/B, pointing to a sensitivity limitation worth monitoring.