(Reuters) – New cancer tests that sequence only a patient’s tumor and not normal tissue could result in a significant number of false positive results, potentially leading doctors to prescribe treatments that might not work, U.S. researchers said on Wednesday.
The findings, published in the journal Science Translational Medicine, call into question the accuracy of increasingly popular tests that look for mutations within tumors that drive cancer growth.
Many laboratories offer such tests to help doctors select personalized cancer therapies. The tests take advantage of new treatments that target changes in the DNA of tumor cells that are important for their survival.
Roche recently paid $1.03 billion for a majority stake in Foundation Medicine, a pioneer in tumor profiling.
The issue is that few of these tests look at DNA from healthy cells to compare which mutations patients were born with and which are unique to the cancer, said Dr. Victor Velculescu of Johns Hopkins and a principal in Personal Genome Diagnostics, a company co-founded by the researchers.
Velculescu and colleagues sequenced tumors and healthy tissue from 815 individuals with a variety of cancers. Two-thirds of mutations in the tumors were also present in healthy tissue.
“They were false positives,” Velculescu said.
The researchers also looked at changes in “actionable genes” – genes for which a treatment is available or being studied in clinical trials – and found nearly half were also present in healthy tissue, suggesting they were not driving the cancer.
Velculescu said the risk of not analyzing both normal and tumor tissue is that patients could be given costly drugs that do nothing to fight their cancer.
Phil Stephens, Foundation Medicine’s chief scientific officer, rejected the assertion that companies need to sequence both the tumor and healthy tissue to get accurate results.
Stephens said most of the false positive variants in the study are for variants of unknown significance. He said his company’s test focuses on genes that are “unambiguously cancer drivers.”
Kenna Shaw of MD Anderson Cancer Center defended the findings, saying there is a risk that some mutations will not be identified in tumor-only sequencing.
“We’re definitely missing things,” she said, adding that the risk is “relatively small.”
Shaw said often companies look at “hotspot” mutations that are clear cancer drivers and these can be found in tumor-only tests. But she said there is no question that having both tumor and normal tissue produces better-quality results. (Reporting by Julie Steenhuysen; Editing by Jonathan Oatis)