Can AI explain my NGS report?
AI can explain the words in a next-generation sequencing (NGS) report, such as variant, VUS, TMB and MSI, and help you list questions. It cannot yet decide which treatment fits your gene changes. In a 2023 JAMA Network Open study, four AI chatbots suggested treatments for fictional cases, and a molecular tumour board found their suggestions fell well short of a human expert's. Treatment choices from an NGS report need your oncologist. Using AI the right way can make your appointment more useful.
What is in an NGS report
- Variants: gene changes found in the tumour, often sorted into four tiers, from strong clinical significance to benign.
- VUS: a variant of unknown significance, which usually should not change treatment.
- TMB and MSI: markers that can make some patients eligible for immunotherapy, whatever the cancer type.
- Evidence levels: databases such as OncoKB grade how strongly a gene change is linked to a drug.
The limits
AI studies on NGS reports mostly use fictional cases, and none shows better outcomes. Experts say AI needs close supervision by a molecular tumour board. Some AI answers also miss new approvals, so always check the date of the evidence. Researchers are now testing expert-guided systems that look up references and learn from tumour board feedback, such as MEREDITH in Germany. Early results improved agreement with clinicians, but these systems are tested on fictional cases and still need a molecular tumour board to check every suggestion.
Questions to ask your doctor
Take these to your next appointment:
- Which gene changes on my report matter for treatment?
- Is there a targeted therapy or immunotherapy that fits these results?
- Can a molecular tumour board review my report?
- Do any clinical trials match my gene changes?
Where OncoVOIS fits
OncoVOIS, a CancerFax cancer-navigation app, explains NGS terms in plain words and helps you prepare questions about targeted therapy and trials. It is available on Google Play for Android phones, and the CancerFax team can help with expert opinions and trials.
Sources
1. Li MM, et al. Standards and guidelines for the interpretation and reporting of sequence variants in cancer (AMP/ASCO/CAP). Journal of Molecular Diagnostics, 2017. https://doi.org/10.1016/j.jmoldx.2016.10.002 2. Benary M, et al. Leveraging large language models for decision support in personalized oncology. JAMA Network Open, 2023. https://pmc.ncbi.nlm.nih.gov/articles/PMC10656647 3. National Cancer Institute. FDA approves pembrolizumab for tumors with high mutational burden, 2020. https://www.cancer.gov/news-events/cancer-currents-blog/2020/fda-pembrolizumab-tmb-approval-genomic-testing 4. Memorial Sloan Kettering. FDA recognizes MSK's OncoKB database, 2021. https://www.mskcc.org/news/milestone-precision-oncology-fda-gives-green-light-msk-s-genetic-database 5. Lammert J, et al. Expert-guided large language models for clinical decision support in precision oncology. JCO Precision Oncology, 2024. https://doi.org/10.1200/PO-24-00478