PS14: Statistical Innovations in Cancer Diagnostics Development and Validation

Cancer diagnostics is a fast-evolving industry fueled by new lab technologies such as liquid biopsy and new computational tools such as artificial intelligence (AI). The technological advancements have brought challenges and opportunities for statisticians in this field, some of which are unique to cancer diagnostics and less well known to other fields (e.g. drug development). This session aims to provide a forum for statisticians in cancer diagnostics to present their innovative works. We aim to cover the following topics in this session: Collaborations in the AI/ML era. For example, a challenge in liquid biopsy is that tumor signal is often confounded by non-tumor signals (e.g. clonal hematopoiesis or germline alterations). To solve the problem, a machine learning classifier was developed and launched on a liquid biopsy platform. Statisticians played a key role in the development, validation and clinical evidence generation for this new biomarker. Speaker: Chang Xu (Foundation Medicine) or colleague Innovative study designs. Clinical trials intended to evaluate therapeutic efficacy in a biomarker positive population need an assay to enroll patients. When a clinical trial assay (CTA) is used for this purpose, a bridging study is required to evaluate concordance between CTA and the companion diagnostic (CDx), and bridge the clinical data from CTA to CDx. Discussion of study designs beyond traditional clinical bridging and non-inferiority will be explored. Speaker: Cui Guo (Foundation Medicine) Recent trend in CDx development and regulation. In practice, multiple laboratory developed tests (LDTs) may be used in trial enrollment to accelerate patient enrollment. In this presentation, we will discuss practical statistical challenges in complex trial and diagnostic study designs, especially recent issues during the communications between drug/device sponsors and the agency. Ongoing FDA/CDRH initiated CDx reclassification will also be discussed. Speaker: Yaji Xu (Johnson & Johnson)