Multi-laboratory study demonstrates high agreement among molecular tumor tests

A recent study of 26 laboratories across the United States revealed high concordance between next-generation sequencing tumor tests. Despite using a variety of sequencing chemistries and instrument platforms, the laboratories returned similar results. 

The Molecular Characterization Laboratory at the Frederick National Laboratory for Cancer Research served as the central laboratory for the study. The other laboratories were a mix of 11 commercial and 14 academic laboratories that run routine next-generation tests for the clinical management of cancer patients.  

“This study shows standard-of-care next-generation sequencing tests performed all throughout the country have high agreement with one another,” said Chris Karlovich, Ph.D., director of Molecular Characterization Laboratory and the senior author on the paper. “You can rest assured that the results that you get in one lab are going to be reproducible in another lab somewhere else, and that is a very important finding.” 

These results were published in Clinical Cancer Research earlier this year. 

Expanding research 

The concordance study arose from a need to support the National Cancer Institute’s Molecular Analysis for Therapy Choice (MATCH) study, one of the largest oncology precision medicine trials. For MATCH, cancer patients were assigned to one of 38 different treatment arms based on the specific mutations found in their tumor. Between 2015 and 2017, MATCH screened around 6,000 patients; however, because of the rarity of some of the mutations, 19 treatment arms remained open. 

To identify more eligible patients, the MATCH study team created a Designated Laboratory Network to allow laboratories across the country to participate. Before being permitted to join the network, the laboratories underwent a rigorous vetting process, which included the concordance study. To participate in MATCH, each laboratory was required to have at least 80% concordance with the Molecular Characterization Laboratory. 

Testing concordance 

Each laboratory tested DNA from eight cell lines and two clinical tumor specimens. The materials included a variety of mutations: single nucleotide variants which are alterations in a single base position in a DNA sequence; insertions and deletions (indels) of bases in the genome; and copy number variants, which are structural genomic alterations that arise due to duplications or deletions of a DNA segment. 

Unexpected results 

For single nucleotide variants and indels, all the labs exceed 88% agreement with the Molecular Characterization Laboratory. For copy number variants there was also high concordance; all but two laboratories exceeded 80% agreement. 

Karlovich explained many of the observed differences were introduced during the bioinformatics analysis. For some laboratories, the assay detected the variant, and it appeared in the raw data; however, the data analysis pipeline filtered it out.  

“The results surprised even me,” said Karlovich. “I knew here was going to be some difference in the way labs filter data, but I thought there would be more difference in labs just not being able to detect some variants because of challenging genomic context.” 

A model network 

Karlovich noted the network and study represent an effective collaborative effort. Many of the laboratories shared confidential data from their validation reports and data analysis pipelines so the investigators could better understand the results and identify the source of any discrepancies. 

“The Designated Lab Network has often been cited as a model for how to screen for rare biomarkers in a clinical trial,” said Karlovich.  

The National Cancer Institute is now leveraging the Designated Laboratory Network, which has expanded to 39 member laboratories, to support ComboMATCH, a large precision oncology study and a successor to MATCH. The new members also underwent concordance testing with the Frederick laboratory before being invited to join the study.