Abstract
PURPOSE: Next-generation sequencing assays for ctDNA analysis are routinely used in the care of patients with advanced non-small cell lung cancer. However, variable assay sensitivities in detection of fusions have been reported. Here, we report on the performance of detecting RET rearrangements in plasma across three commercial next-generation sequencing laboratories. EXPERIMENTAL DESIGN: Banked plasma from the phase 3 LIBRETTO-431 trial was studied. For each patient (n = 60) with a known RET fusion by local tumor tissue genotyping, pretreatment plasma was divided into two 3-mL aliquots and tested on two of three: Guardant Health's Guardant360, Foundation Medicine's FoundationOneLiquid CDx, and Resolution Bioscience's ctDx-First. A round-robin comparison was performed across vendors using three pairwise comparisons of 20 patients each. On an exploratory basis, agreement of fusion breakpoint calling between plasma and tissue and determinants of false negatives in plasma were assessed. RESULTS: Of 40 samples received by each laboratory, 100% (40/40), 92.5% (37/40), and 90% (36/40) were successfully sequenced by Guardant360, FoundationOne Liquid CDx, and ctDx-First, with a RET fusion or rearrangement detected in 60% (24/40), 63.9% (23/36), and 67.6% (25/37) of cases, respectively. Discordant results included rare and common RET translocations but were usually below allelic frequency of 0.5%. Of samples with a RET fusion detected in plasma and a reported fusion partner by tumor assay, the same fusion partner was identified in tissue and liquid 81% to 89% of the time. CONCLUSIONS: Our results support the utility of ctDNA assays concurrently with tissue testing for detection of translocations, with opportunities to further optimize performance. See related commentary by Davies, p. 2264.