Abstract
Abstract Lynch syndrome, the most common cause of hereditary colorectal cancer, is due to pathogenic germline variants in mismatch repair (MMR) genes. Variants of uncertain significance (VUS) in MMR genes hinder diagnoses and clinical management. Multiplex assays of variant effects (MAVEs) are high-throughput cellular assays designed to provide functional data on variant activity. Successful MAVE experiments accurately define the pathogenicity of single nucleotide variants and small insertions-deletions which can then be applied to clinical variant classification (VC). Here we report internally generated MAVE datasets for three MMR genes (MSH2, MLH1 and PMS2) and estimate their impact on patients. Three separate pools of targeted variants were generated for three MMR genes. Pooled variants were introduced into cell populations such that each cell contained one variant copy of an MMR gene. Single-cell RNA expression profiles were generated for the cells which were then normalized and filtered. Supervised machine learning was used to find patterns in gene expression that were different between cells harboring established pathogenic (P) and benign (B) variants and to assess the accuracy of variant predictions. VC impact was estimated for all datasets by simulating VCs with the added MAVE evidence compared to the original VCs. Variants with an original VC that was different after MAVE evidence were counted as impacted. Overall results are shown in Table 1, including the number of variants successfully targeted, the number of known P and B variants used for training and testing, the area under the receiver operating characteristic curve (AUROC), the distribution of calibrated predictions, and the estimated patient impact for each model. Table 1. Gene MSH2 model MLH1 model PMS2 model Variants successfully targeted (n) 223 212 214 Known pathogenic (P) and benign (B) variants used to train the model 29 P 31 P 14 P 72 B 55 B 71 B Known pathogenic (P) and benign (B) variants used to test the model 14 P 14 P 7 P 32 B 22 B 31 B AUROC 0.9996 0.9642 0.9918 # of variants strongly predicted P (PPV ≥ 95%) or B (NPV≥ 95%) 44 P 52 P 26 P 177 B 160 B 183 B # of variants with uncertain prediction (PPV < 80%, NPV < 80%) 2 0 5 Estimated # of impacted variants 4 LP>P 0 LP>P 4 LP>P 3 VUS>LP 5 VUS>LP 1 VUS>LP 26 VUS>LB 17 VUS>LB 17 VUS>LB 31 LB>B 7 LB>B 6 LB>B The MSH2, MLH1, and PMS2 MAVE models provide a high-quality source of functional evidence for VC with the potential to impact many patients in the future. Citation Format: Wolfgang Michael Korn, Samskruthi Padigepati, David Stafford, Flavia Facio, Britt Johnson, Keith Nykamp, Jason Reuter. Novel MAVE models for MLH1, MSH2, and PMS2 have high accuracy [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 7329.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.