Prediction of oncogene mutation status in non-small cell lung cancer: A systematic review and meta-analysis with a special focus on artificial-intelligence-based methods

BackgroundIn non-small cell lung cancer (NSCLC), alternative strategies to determine patient oncogene mutation status are essential to overcome some of the drawbacks associated with current methods. We aimed to review the use of radiomics alone or in combination with clinical data and to evaluate the performance of artificial intelligence (AI)-based models on the prediction of oncogene mutation status. MethodsA PRISMA-compliant literature review was conducted. The Medline (via Pubmed), Embase, and Cochrane Library databases were searched for studies published through June 30, 2023 predicting oncogene mutation status in patients with NSCLC using radiomics. Independent meta-analyses evaluating the performance of AI-based models developed with radiomics features or with a combination of radiomics features plus clinical data for the prediction of different oncogenic driver mutations were performed. A meta-regression to analyze the influence of methodological/clinical factors was also conducted. ResultsOut of the 615 studies identified, 89 evaluating models for the prediction of epidermal growth factor-1 (EGFR), anaplastic lymphoma kinase (ALK), and Kirsten rat sarcoma virus (KRAS) mutations were included in the systematic review. A total of 38 met the inclusion criteria for the meta-analyses. The AI algorithms sensitivity/false positive rate (FPR) in predicting EGFR, ALK, and KRAS mutations using radiomics-based models was 0.753 (95% CI 0.721-0.783)/0.346 (95% CI 0.305-0.390), 0.754 (95% CI 0.639-0.841)/ 0.225 (95% CI 0.163-0.302), and 0.744 (95% CI 0.605-0.846)/0.376 (95% CI 0.274-0.491), respectively. A meta-analysis of combined models was only possible for EGFR mutation, revealing a sensitivity/FPR of 0.800 (95% CI 0.767-0.830)/0.335 (95% CI 0.279-0.396). No statistically significant results were obtained in the meta-regression. ConclusionsRadiomics-based models may represent valuable non-invasive tools for the determination of oncogene mutation status in NSCLC. Further investigation is required to analyze whether clinical data might boost their performance.