DDTBMQ000011: Advance tumor volume change as measured by CT as a pharmacodynamic biomarker for new oncologic drug clinical trial therapy of solid tumors to Stage 2 of FDA biomarker qualification

点击翻译按钮获取中文摘要

Project Summary/Abstract The overall objective of this project is to advance the biomarker qualification application DDTBMQ000011 from the current Stage 1 to Stage 2 of the FDA biomarker qualification process. The scope of this project is based on recommendations from the FDA BQP review team that were communicated in the Determination Letter for accepting the legacy biomarker qualification proposal into the 507 Process, on additional advice received from the subsequent debriefing meeting, and on FDA guidance documents. A FDA BQP-suggested study will be conducted to assess agreement between the QIBA Profile’s CT volumetry biomarker definition of change (CTvol) and the RECIST system using the categories of responsive, stable, or progressive disease. Disagreements between the two approaches will be further broken down as: (i) substantive disagreement, where the disagreement between RECIST and CT volumetry cannot be attributable to improved sensitivity with CT volumetry and (ii) disagreement potentially due to improved sensitivity with CT volumetry. Factors predictive of disagreement will be identified. The null hypothesis is that the proportion of lesions with substantive disagreement is >15%; the alternative hypothesis is that the proportion of lesions with substantive disagreement is <15%. Images will be retrospectively collected from prior studies of subjects undergoing serial CT imaging in Phase 3 drug trials. Three readers will be recruited with very different levels of experience to participate in a prospective reader study of 234 cases. Study readers will measure each target lesion on the baseline and follow-up scans using two image analysis software tools. Both unidimensional and volume measurements of each target lesion at each timepoint will be taken and recorded for each case. From these data, measurements of change will be constructed when the same reader makes the measurements at each time point, as well as when different readers make the measurement at baseline and follow-up. The estimates of change will then be classified into three categories (partial or complete response, stable, or progression) using the two approaches (RECIST and QIBA Profile claim). The disagreement in classification between the two approaches will be reported. Multiple-variable logistic regression models will be fit to assess contribution of lesion characteristics (size, shape, location), magnitude of change as measured by CTvol, scanner model, image analysis software, whether imaging was QIBA-Profile-conformant or not, and reader protocol (same or different reader at two time points). The proposed work has great potential to validate the reproducibility of an emerging quantitative imaging biomarker (i.e., CT volumetry) to more precisely determine tumor response or progression in therapy trials.