AI oncology algorithm-based prioritisation of EGFR inhibitors in case of rare EGFR mutations

Abstract

Abstract Background Around 30% of non-small cell lung cancer (NSCLC) patients harbor sensitizing mutations in the epidermal growth factor receptor (EGFR) gene. Most common EGFR mutations are exon 19 deletions and EGFR-L858R, that comprise 85% of EGFR-mutated cases, and whose drug sensitivity profile is already known. However, rare EGFR mutations represent 10-15% of EGFR-mutated NSCLCs with diverse sensitivity to different EGFR inhibitors. Methods Scientific literature and database search were conducted to identify pathogenicity and drug sensitivity of rare and common EGFR mutations. Structured data were uploaded in the database of the RealTime Oncology Treatment CalculatorTM (Calculator) to enable automated interpretation of variants. A structured database of clinical trials testing EGFR targeted therapies was also established for matching trial search. Molecular diagnostics of an NSCLC patient were performed (next-generation sequencing of 58 genes and FISH analyses) for therapeutic decision support. A rare EGFR mutation was identified (H773_V774insNPH) and automated classification was carried out using the Calculator. Results The Calculator is now able to automatically classify the most frequent 100+ EGFR mutations based on structured scientific evidence and by using a proprietary algorithm it prioritizes the associated EGFR inhibitors. The system can also handle complex molecular profiles so that resistance-associated alterations (e.g. KRAS mutation or ALK translocation) can be identified and optimal targeted therapies can be suggested. Matching clinical trials can be listed within seconds based on histology, biomarker criteria, compounds, previous therapies, etc. The NSCLC patient with the rare EGFR mutation (who had quickly progressed on gefitinib) had stable disease for 2 years on afatinib, which compound was suggested by the Calculator. Gefitinib was negatively associated with the molecular profile. Conclusions This case clearly demonstrates that knowledge on the drug sensitivity profile of the exact mutation is very important for the therapeutic decision. We demonstrate that RealTime Oncology Treatment CalculatorTM is an efficient tool for evidence-based prioritization of EGFR inhibitors even in the case of rare EGFR variants. Legal entity responsible for the study Oncompass Medicine Hungary Ltd. Funding National Research, Development and Innovation Office, National Oncogenomics and Precision Oncotherapy Program (NVKP_16-1-2016-0005), WESPED (KFI_16-1-2016-0048). Disclosure All authors have declared no conflicts of interest.