Discordant Results of Next Gen Tumor Profiling Raise Concerns

Conversations regarding oversight of diagnostic testing focus on the clinical and analytical validity of such testing and the ability to replicate results reliably. For example, the U.S. Food and Drug Administration's attempts to develop an oversight framework for laboratory developed testing focus on the reliability of these locally developed tests.

New technology and test methods raise the same concerns. Theranos first began facing troubles when reports claimed test results from Theranos' finger- stick testing methods differed from tests run using traditional testing platforms and methods.

A new study raises similar reliability questions based on discordant results involving next-generation sequencing (NGS), which has been touted as having significant promise. NGS is increasingly becoming commonplace, for example, to match cancer-associated alterations with targeted treatments. The FDA has held at least one workshop devoted to discussing potential standards for evaluating NGS and evidence of clinical performance and reliability. (See "FDA Seeks Stakeholder Feedback on Evaluating Next-Generation Sequencing Tests," NIR, March 24, 2015).

This study found that marked differences in results between two commercially available genetic tests for oncology patients may be "clinically relevant."

The research letter, published online Dec. 15, 2016 in JAMA Oncology, compared the tissue-based FoundationOne test (F1; Foundation Medicine) with the blood-based Guardant360 (G360; Guardant Health) test. F1 characterizes the exons of 315 cancer-associated genes and introns from 28 genes involved in rearrangements, while G360 sequences 70 genes from cell-free circulating DNA. Previous published studies have shown that both the F1 and G360 tests have high specificities (above 99 percent), but lower sensitivities.

The present study compared results from both tests in nine patients seen at a community oncology practice. The level of concordance between the platforms was compared among the two men and seven women (mean age, 61 years). Testing occurred from April 14, 2015 to Jan. 30, 2016. In addition to comparing identified genomic alterations, test results were compared regarding recommended drugs.

The researchers found one patient had no identified genetic alterations using either test. Among the remaining eight patients, 45 alterations were identified, but only 10 alterations (22 percent) were concordant between the platforms. For two of the eight patients, there was no concordance among the reported alterations. Alterations that are unique to the F1 test, which detects a "much broader range" of aberrations than G360, were excluded from analysis. Concordance improved "only slightly" to 28 percent (5 of 18 alterations) when comparisons were limited to variant allele frequencies of 1 percent or greater.

For the eight patients with identified alterations, 36 drugs were recommended, in total. However, only one-quarter of the drugs were recommended for the same patients by both platforms. In five patients there was no overlap between the drugs recommended by the two tests. Concordance among recommended drugs improved to 62 percent (8 of 13 drugs), when reported mutations were also concordant.

In seeking an explanation for the discordant test results, the authors cite differences in timing between the two tests as a possible source, but note that seven of the eight patients with reported alterations underwent both tests within a 2.5-month period. Other potential sources of the discordance are tumor heterogeneity and differences in the variant-interpretation process.

"Since both the F1 and the G360 tests are performed in thousands of patients with cancer each year, these findings are clinically relevant," write the authors led by Nicole M. Kuderer, M.D., from University of Washington, Seattle. "In-depth comparisons of next-generation sequencing tests across larger numbers of patients with cancer are needed to improve concordance and clinical utility."

The authors note that theirs was not the first to identify "significant discordance." Two studies comparing tissue-based next-generation sequencing tests and another report also comparing the F1 and G360 tests, all found discordant test results.

Takeaway: Initial comparisons of commercially available nextgeneration sequencing tests to identify cancer-related variants targetable by therapies, raise concerns regarding the potential clinical relevance of discordant genetic findings and resulting drug recommendations.