NEW YORK (GenomeWeb) – Tests from two next-generation sequencing companies that examined samples from the same nine cancer patients didn't always agree on what genetic alterations the patients' tumors harbored or what treatments should be used.
A team of researchers led by the University of Washington's Tony Blau examined reports that nine patients received after taking Foundation Medicine's FoundationOne test and Guardant Health's Guardant360 test. As Blau and his colleagues reported in a preliminary study appearing in JAMA Oncology today, they found that only 22 percent of alterations were reported back by both companies. While the researchers noted that one test likely uncovers a broader range of changes, they actually suspected that the discrepancies are instead due to tumor heterogeneity and differences in how the companies decide which variants to report.
"Our findings indicate that the output from genetic testing can differ markedly depending on which test is applied," Blau and his colleagues wrote in their paper.
While both the FoundationOne and Guardant360 tests sequence clinical cancer samples, they go about it a little differently. The FoundationOne test sequences the exons of 315 cancer-linked genes and the introns of 28 genes involved in rearrangements from tissue samples. The Guardant360 test, meanwhile, sequences some 70 genes from cell-free circulating DNA found in the blood. Both tests, the researchers noted, have greater than 99 percent specificities, but slightly lower sensitivities.
Blau and his colleagues recruited nine patients who underwent testing on both platforms to their study. Five of the patients had breast cancer, and the others had pancreatic cancer, thymic carcinoma, lung cancer, and salivary gland cancer.
Neither test uncovered any genetic alterations for the patient with thymic carcinoma. The tests uncovered a total of 45 mutations for the other patients, but only 10 of those mutations, or 22 percent, were found by both platforms. For two of these eight patients, there was no overlap between mutations found by the tests.
The tests also recommend drugs that might benefit the patients based on their results. All in all, the two tests recommended 36 drugs, but only nine drugs were recommended for the same patients by the different platforms, and five patients have no overlap in the drugs they were recommended. However, when the researchers restricted that part of their analysis to patients for whom the tests reported the same mutations, the concordance among the drugs recommended increased to 62 percent.
Blau and his colleagues offered a number of potential causes for the inconsistencies they observed between the tests. The FoundationOne test examines a wider range of possible mutations than the Guardant360 test, but the researchers noted that they didn't include variants that would have only been caught by the FoundationOne test. Another possible cause, they said, could be the timing when the patients were tested, but they seven of the eight patients took both tests within a 10-week time period.
That leaves tumor heterogeneity, differences in how the companies decide to return variants, and, as both companies also noted, the different samples the tests rely upon.
In the paper, the researchers noted that the Guardant360 test did often return low-frequency variants. When the researchers constrained their analysis to only variants found through Guardant360 testing at a higher than 1 percent frequency, some of the discordance between the tests disappeared.
Foundation Medicine Chief Medical Officer Vince Miller argued that tumors aren't as heterogeneous as thought. He said that samples from different spots of the same tumor are often highly concordant and that a single biopsy can capture, in most cases, the key alterations. He added that part of the discordance might be because tumors don't shed cells at a constant rate to be caught by a liquid biopsy approach.
He added that results like this suggest to him to "stick with what's tried and true" and to rely on tissue samples when they are available.
Conversely, Justin Odegaard, the senior medical director and laboratory director at Guardant Health, argued the less-frequent variants caught by liquid biopsy represent heterogeneity within the solid tumor or even metastases.
Overall, to Odegaard, the study suggested that the approaches might be complementary.
"I think it's important to note that both techniques found things that the other didn't," Odegaard said. "And that gives you two shots on goal for finding something that can really get that patient on the best therapy possible."