Detection of microsatellite instability (MSI) is recommended for all patients with colorectal cancer (CRC). Current clinical reference methods are immunohistochemical (IHC) staining of mismatch repair (MMR) proteins and/or PCR analysis of frequently mutated repetitive regions of DNA. The prototype Idylla™ MSI Test has been developed using a new set of short homopolymers located in the ACVR2A, BTBD7, DIDO1, MRE11, RYR3, SEC31A & SULF2 genes. This marker set allows probe-based detection with great specificity in a simplified workflow compared to current methods.
Repeat length with this set of biomarkers was determined on 333 formalin-fixed and paraffin-embedded (FFPE) CRC samples using Idylla™ MSI Test prototype cartridges, which allow a fully automated workflow including sample preparation, DNA amplification and automated repeat length calling. A neural network based algorithm was built on a large cohort of reference/patients samples (n > 3000) obtained from different clinical sites (n > 10) and different ethnic groups (n = 5). Three-hundred fourteen samples were characterized by means of the Promega MSI analysis system and 272 samples by means of MMR protein IHC staining. Approximately 30% of the samples included in the study were previously characterized to be MSI-H by either one of these methods.
Concordance analysis revealed an overall agreement of 98.7% (96.7%-99.5% 95% CI) with Promega and 97.6% (94.8%-98.9% 95% CI) with IHC analysis. Analysis of consecutive sections of 182 samples with the three methodologies revealed a higher number of invalid results for Promega (3.8%) and IHC (13.2%) compared to the prototype Idylla™ MSI Test (2.2%).
This study verified the robustness of the prototype Idylla™ MSI Test including novel MSI biomarkers to discriminate MSI-H from MSS status on a large and diverse set of CRC samples. The study was conducted in multiple centers demonstrating the possibility of a rapid and fully automated analysis for MSI testing close to the point of need. The prototype Idylla™ MSI Test provided accurate and reliable results within 150 minutes from just one FFPE tumor section (no normal reference sample required).
Clinical trial identification
Legal entity responsible for the study
B. De Craene, J. Van de Velde, E. Bellon, M. Gazin, E. Rondelez, L. Vandenbroeck, T. Vanhoey, N. Elsen, K. Decanniere, E. Sablon, G.G. Maertens: Employee: Biocartis NV. E. Watkin: Advisory board, speaker’s fees: MSD. All other authors have declared no conflicts of interest.