Meenakshi Mehrotra
Meenakshi Mehrotra Ph.D is an Assistant Professor in Department of Pathology, Molecular and Cell-Based Medicine Icahn School of Medicine at Mount Sinai, New York NY. She is also an Assistant Director of Clinical Molecular Pathology Lab, Mount Sinai Health System. NY. She received her doctoral degree in Molecular Biology and genetic Engineering from the University of Lucknow, INDIA. Dr. Mehrotra did her postdoctoral training in at the University of Texas MD Anderson cancer Center Houston, Tx and has an excellent background in Clinical genomics and Molecular Diagnostics. Dr. Mehrotra is involve in new assay and technology development, which involves somatic mutation detection, identification of translocations, copy number detection for cancer-related clinical molecular diagnostics and research. She has expertise in different next generation sequencing platforms and played a key role in the validation and implementation of a next generation sequencing assay for the detection of somatic mutations in solid tumors and liquid biopsies. She participates in clinical and translational studies in oncology and her areas of expertise include evaluation and validation of new diagnostic platforms for applications in oncology which can be used for a variety of clinical and investigational applications.
Abstract
Meenakshi Mehrotraa, William Lamb, Evelyne Kouameb, Michelle Guanb, Brett Baskovicha, Matthew Crokena, Jane Houldswortha
aIcahn School of Medicine at Mount Sinai, New York, NY, United States; bMount Siani Health System, New York, NY, United States
Technical barriers, panel complexity and bioinformatics are major challenges for the clinical implementation of large NGS panels. For optimization of the 500+ gene Oncomine comprehensive plus targeted amplicon based NGS assay to detect SNVs, Indels, CNVs, fusions and TMB, 40ng DNA and RNA from 115 specimens from different tumor types (tumor cellularity ≥10%) were used for library preparation and sequenced on the Ion S5XL sequencer. Ion Reporter 5.18 (IR) and Alissa custom classification trees were optimized for known variants. CDKN2A, PTEN and TERT genes were underperformers. SNVs/Indels demonstrated 95% concordance with OCAv3 panel (n=71) sequenced comparator specimens, and 94% and 56% for SNVs and Indels respectively compared to a hybrid capture 324 gene panel (n=22). Discordant SNVs had poor coverage (<100x) and low VAF (<3%), while 76% discrepant Indel calls were in homopolymer region of length ≥5. After augmentation of the CNV IR baseline with 44 tumor and normal specimens, concordance for 85 CNVs was 92% for known CNV ≥6 copies, 83% for ≥5-6, and 46% for ≥4-5, and 100% for loss. Fusions were detected with 100% concordance. 22 specimens with prior TMB score (14 high (≥10) and 8 low) demonstrated 86% concordance. TMB discrepant specimens (range 11-25) showed low score due to missed variants in homopolymer region. In the current optimization study, challenges faced in clinical implementation included lower confidence for Indel calling in homopolymer length >4, CNV gain 4-5 and borderline TMB.