Statistical Innovation in the Application of ctDNA in Oncology Drug Development
Chair: Shibing Deng (Pfizer)
Speaker: Hong Zhang, PhD (Pfizer)
Title: Quantifying the Benefits of ctDNA-Guided Designs for Clinical Trials with Time-to-Event Endpoints
Abstract: In the era of precision medicine, many biomarkers are found to impact patient prognosis, risk assessment, and disease progression. A wide spectrum of novel designs has been proposed to utilize the prognostic and predictive biomarkers to increase the efficiency of clinical trial designs compared to traditional designs. However, the advantages of biomarker guided designs are often demonstrated without quantitatively accounting for the slower enrollment rate of biomarker positive patients. We extend the study duration model in Machida et al. (Statistics in Medicine, 2021) to be a general theoretical framework with mixture distributions incorporating heterogeneous patient population. Extensive simulations are performed to understand how the dynamics of the disease and biomarker characteristics could affect the study duration. Several influential parameters are identified and their impacts are discussed. Examples include patient median survival, enrollment rate, biomarker prevalence and effect size. Re-assessments of multiple past trials are conducted to validate the prediction accuracy and to evaluate circulating tumor DNA (ctDNA) as a promising candidate for biomarker-guided trials. The results provide quantitative evidence for adopting biomarker guided designs that may reduce the sample size and/or shorten the study duration and increase the statistical power.
Speaker: ChangHong Song, PhD (FDA)
Title: Statistical Innovation for ctDNA Assays and Their Application in Oncology Drug Development
Abstract: Circulating tumor DNA (ctDNA) are getting more and more applications in many different areas. For precision medicine, ctDNA based companion diagnostic devices have been approved by the U.S. Food and Drug Administration to help identify subjects that can benefit from a therapeutic product. In oncology, ctDNA has been used for clinical trial enrichment and stratification. ctDNA has also been considered a promising biomarker that has the potential to be used as an early endpoint for clinical trials, monitor treatment response, etc. This presentation discusses statistical innovations and opportunities for ctDNA assay evaluations and the application of ctDNA assays in Oncology Drug Development.
Speaker: Hillary Andrews, PhD (FOCR)
Title: The Friends’ ctMoniTR Project: Aggregate Patient Level Analyses to Understand Associations Between ctDNA and Long-term Outcome
Abstract: Circulating tumor DNA (ctDNA) holds promise as an early endpoint to predict long-term treatment outcomes. However, patient- and trial-level association measures are needed to support its use in clinical trials and regulatory decision-making. Friends of Cancer Research (Friends) has an ongoing collaboration with industry, academic, and U.S. government stakeholders to assess associations between change in ctDNA levels and long-term outcomes in aggregate patient-level datasets. Industry sponsors donate patient-level clinical trial data and our third-party data aggregator and statistical group, Cancer Research and Biostatistics (CRAB), performs analyses. The full group of stakeholders then discusses findings and recommendations for the field. To date, Friends has overseen two analyses – one in patients with advanced non-small cell lung cancer (aNSCLC) treated with an anti-PD(L)1 and one in patients with aNSCLC treated with a tyrosine kinase inhibitor. In both cases, decreases in ctDNA levels from baseline to an on treatment timepoint associated with improved long-term outcomes. Ongoing collaborative work focuses on a new dataset of patients with aNSCLC who were treated with either an anti-PD(L)1 or chemotherapy with ctDNA collected at multiple timepoints. This dataset affords the opportunity to interrogate dynamics of ctDNA levels over time and to further understand another treatment modality. Collaborative approaches to data analysis are critical to expedite our understanding of how changes in ctDNA levels associate with long-term outcomes.
