Matched tumor-normal pairs are necessary to reliably identify and interpret mutations found in patients' tumors to guide treatment, according to researchers from Johns Hopkins University School of Medicine and Personal Genome Diagnostics.
However, as Julia Karow writes at GenomeWeb, many diagnostic labs only sequence tumor tissue and not matched normal tissue, often for reasons of cost.
As Hopkins' Victor Velculescu and his colleagues report in Science Translational Medicine this week, they sequenced 815 tumor-normal samples representing 15 tumor types, focusing on 111 clinically relevant genes. From this, they found that more than three-quarters of cases harbored somatic changes in genes associated with known therapies or with ones undergoing clinical trial testing.
When they compared the tumor-normal matched and tumor-only analysis approaches, they found that about a third of the mutations identified by the tumor-only approach were false positives. And, after accounting for multiple mutations, nearly half of the patients were affected by such false positives, the researchers say.
"We were surprised by the magnitude of the errors that can come out of this type of analysis," Velculescu tells the Wall Street Journal. "It would be a shame to use these targeted therapies and not use the right ones."
However, cancer diagnostic lab Foundation Medicine's Phil Stephens tells the Journal that the study is "naïve" and potentially "misleading." He adds that "no responsible diagnostic company would ever report out" the mutations the Hopkins researchers found to be false positives.
MD Anderson's Kenna Shaw adds that "it's an overreach to suggest that we're misdirecting so many patients" by not also sequencing normal tissue. She also says that mutations whose cancer role is unknown are not typically reported back to clinicians as a drug target.
Still, as Karow writes, some institutes like Memorial Sloan-Kettering Cancer Center are moving toward matched tumor-normal analysis.