An analysis of two commercially available, next-generation platforms in the same cancer patients yielded very different results, calling into question the reliability of these tests in directing treatment options.
Researchers from the University of Washington in Seattle analyzed two widely used gene sequencing tests, FoundationOne from Foundation Medicine, and Guardant360 from Guardant Health. A research letter detailing the results was published in JAMA Oncology.
Cancers arise from collections of mutations. Both tests seek to characterize the mutations that exist within a patient’s cancer, explained C. Anthony Blau, MD, a professor of medicine at the University of Washington School of Medicine’s Division of Hematology, and the study’s lead author. The FoundationOne test extracts DNA from tumor tissue to characterize the mutations within that sample, he told CLN Stat.
Guardant bases its technology on the observation that DNA from various tissues—including cancer tissues—sheds into the bloodstream. The test therefore “extracts DNA from the cell-free fraction of the blood to characterize mutations,” said Blau, who is also the director of the university’s Center for Cancer Innovation.
Both tests yield high specificities, although they tend to score lower on sensitivities. “Little is known about how different next-generation sequencing tests compare when used in the same patients with cancer,” the researchers commented in the research letter. In this study, they focused just on aberrations that the FoundationOne and Guardant360 tests were able to detect.
Analyzing the test reports among nine patients from a community cancer care practice, researchers discovered inconsistent results among the two platforms. With the exception of one patient, in which neither test found genetic alterations, the remaining eight had a total of 45 genetic mutations. Only 10 of those mutations were detected by both tests. In the case of two patients, none of the results matched for the two tests.
Of the 36 potential treatment drugs identified for the eight patients with genetic mutations, just nine were listed by both test reports. “For five of the eight patients, there was no overlap in recommended drugs between the two reports,” Blau said. He offered that some, but not all, of these instances were due to recommendations Guardant made, based on low frequency mutations. “Giving a drug that affects only a small fraction of tumor cells is unlikely to be effective,” he observed.
Overall, the findings indicate that output from genetic testing can differ markedly depending on which test is applied, the researchers concluded.
They offered several hypotheses for the disparate results. One possibility is the FoundationOne test detected a wider range of aberrations than the Guardant360 test. “However, the discordance we observed was not due to this group of mutations,” Blau noted.
The Guardant360 test examines cfDNA in blood to characterize genes that frequently mutate in cancer. “When they find a mutation, it is invariably mixed with many instances of the same sequence that’s not mutated,” he explained. Normal cells, and tumor cells that lack the specific mutation being studied (as many mutations occur in just a low fraction of tumor cells), contribute to these nonmutated sequences. The study results suggest that low frequency mutations reported by Guardant and not by FoundationOne, may explain some of the disparities between the tests.
“These low frequency mutations may be false positives, but it is more likely that they represent mutations that occur in only a small fraction of tumor cells,” Blau said.
Variations in timing among the two tests could have led to different results. Yet, the investigators point out that seven of the eight individuals with mutations took these tests within the same time period. “Tumor heterogeneity and differences in the processes used to determine whether a variant is reported are also likely contributors to discordance,” they suggested.
The findings are significant from a clinical perspective, given that thousands of cancer patients use these tests on an annual basis. For this reason, the researchers suggested a follow-up comparison study of the two tests using a larger pool of test subjects.