clinical trials
RESOLVED2
FDA
Prediction of Drug Approval After Phase I Clinical Trials in Oncology: RESOLVED2
Summary
- Challenge in Oncology Drug Development: The field is currently facing an increase in the number of antineoplastic agents (ANAs) entering phase I clinical trials (P1CTs) and a high attrition rate for final FDA approval.
- Objective: Development of a machine learning algorithm, RESOLVED2, to predict drug development outcomes, aiding in early go/no-go decisions post-P1CTs.
- Data Sources: Utilized PubMed abstracts of P1CTs and pharmacologic data from the DrugBank5.0 database.
- Modeling Approach: RESOLVED2 was trained to model FDA approval-free survival using machine learning methods, based on 28 key variables out of 1,411.
- Performance: Achieved a weighted concordance index (IPCW) of 0.89, outperforming a model based on efficacy/toxicity (IPCW, 0.69).
- Results: At 6 years follow-up, 73% of drugs predicted to be approved were approved, and 92% of drugs predicted not to be approved remained unapproved.
- Predictive Accuracy: Drugs predicted to be approved were 16 times more likely to be actually approved than those predicted not to be approved.
- Conclusion: RESOLVED2 accurately predicts time to FDA approval post-P1CTs, demonstrating that machine learning can effectively forecast drug development outcomes.
Drug development in oncology is a fast-evolving field with numerous challenges. More than 1,000 antineoplastic agents (ANAs) were under investigation in 2018. Oncology had the highest overall attrition rate for US Food and Drug Administration (FDA) approval from phase I (95% between 2006 and 2015), phase II (92%), and phase III (67%) trials. The community aims to limit the recruitment of patients to phase II and/or large phase III studies that evaluate treatment that will not be approved for various reasons: It impairs recruitment of patients in other studies, slows down the whole drug development process, and results in substantial financial loss for the pharmaceutical industry and academic institutions. Exposure of patients to ineffective treatments and financial loss has urged the pharmaceutical industry and academic investigators to develop new tools to enhance drug development strategies, such as computer-assisted decisions.
Phase I trials in oncology usually are dedicated to safety analysis and meanwhile can provide early signals of efficacy of the compounds. Classic strategies to improve research and development are the use of surrogate markers of efficacy (overall response rate as a surrogate of overall survival) or predictive biomarkers of efficacy (molecular alterations from the tumor or liquid biopsy).The biomarker-based strategy used in phase I can significantly increase response rate and the likelihood of FDA approval...
Click for Source
clinical trials, RESOLVED2, FDA
