SALT LAKE CITY – Roche has presented data on the concordance of its Avenio Tumor Tissue comprehensive genomic profiling (CGP) assay with alternative products from its Foundation Medicine subsidiary and Illumina.
In a sponsored session Wednesday at this year's annual meeting of the Association for Molecular Pathology, Roche Diagnostics Solutions Medical Affairs Lead Helle Sorensen shared data from studies comparing Avenio with the FoundationOne CDx and Illumina's TruSight Oncology (TSO) 500.
A study of 144 samples that were processed by both the Avenio Tumor Tissue and FoundationOne assays showed a "high degree of alignment" for all genomic variants and signatures, such as genomic loss of heterozygosity (gLOH), microsatellite instability (MSI), and tumor mutational burden (TMB), Sorensen said.
The Avenio CGP kitted assays, launched in the US about a year ago, draw on content found in the FoundationOne assay, which is available as a service. Roche wholly acquired Foundation Medicine in 2018. The TSO 500 is also a kitted assay for CGP of solid tumors.
For the TSO 500 comparison, Roche compared the number of variants categorized as Tier I/II by the European Society for Medical Oncology (ESMO) Scale for Clinical Actionability of Targets (ESCAT). Avenio identified 393 while TS0 500 identified 612, with overlap of 281 variants. Avenio caught 11 copy number alterations while TSO 500 only caught six of those, all ERBB2 amplifications. Overall, Avenio caught fewer variants due to assay design and filtering of "putative benign variants," Sorensen said.
Both comparison studies analyzed separate cohorts of 144 formalin-fixed, paraffin-embedded tissue specimens including prostate, breast, and colon, among other cancers. The TSO 500 comparison study also included lung cancer samples. For the FoundationOne comparison, samples were sent to Foundation Medicine for analysis, while Avenio kits were processed by Signature Diagnostics in Potsdam, Germany.
Overall agreement between the assays was 99.8 percent for short variants, 99.4 percent for complex variants, and 98.3 percent for copy number alterations. For genomic instability metrics, Roche provided the weighted kappa coefficient — a measure of agreement. The coefficient for MSI was "identical" for 45 samples. For TMB, it was 0.88; for gLOH, it was 0.82.
Sorensen also shared a slide containing data from another study comparing Avenio to FoundationOne — first presented in a poster at the ESMO annual meeting last month — showing multiple labs around the world achieving high percent positive agreement (PPA) between the assays. SNVs, rearrangements, TMB, and MSI were all nearly identical across labs and at or close to 100 percent PPA to FoundationOne. For indels, one lab reached 100 percent agreement, while all others reached 98.8 percent agreement. CNAs were more varied, with agreement between 87.9 percent and 96.6 percent, and an average agreement of 92 percent.
Avenio had higher coverage, at 1,870X, compared to coverage of 460X for the FoundationOne assay, Sorensen said.
For the TSO 500 study, Roche compared variants identified by both assays and classified as ESCAT Tier I/II. Roche obtained variant tiers for Avenio with its proprietary Navify software; for TSO 500, it used analysis from Pierian, now doing business as part of Velsera.
For the remainder of the 393 gene variants that Avenio called as ESCAT Tier I/II, TSO 500 called 77 as Tier III variants and made no call for 35.
Using American (AMP/ASCO/CAP) variant classification guidelines, TSO made one variant call as Tier I that Avenio did not identify, a KRAS G12F mutation. This was "due to being slightly below the 5 percent variant allele fraction cutoff," Sorensen said.