Comments on design and sequence of biomarker studies Constantine Gatsonis Center for Statistical Sciences Brown University ACRIN Fall Meeting, 2008 ACRIN EISC 2008 Gatsonis, Brown University
Clinical use vs use as trial endpoint Before Tx: stratification in order to decide on alternative treatments During Tx: early prediction of treatment response basis for modifying therapy After Tx: monitoring for recurrence ACRIN EISC 2008 Gatsonis, Brown University
Clinical use vs use as trial endpoint Use as Phase II trial endpoint Does image-based marker predict Phase III endpoints at the patient level at the trial level Combining information from a group of studies is needed to “validate” marker for endpoint use. ACRIN EISC 2008 Gatsonis, Brown University
Aspects of marker performance Assay validation Reliability/ Reproducibility Diagnostic/ predictive accuracy Effect on clinical outcomes For clinical practice X x For Phase II endpoint ACRIN EISC 2008 Gatsonis, Brown University
Aspects of marker performance Reliability/ Reproducibility Diagnostic/ predictive accuracy Effect on clinical outcomes For clinical practice X x For Phase II endpoint More evidence is needed for Phase II endpoint validation ACRIN EISC 2008 Gatsonis, Brown University
Aspects of marker performance Reliability/ Reproducibility Diagnostic/ predictive accuracy Effect on clinical outcomes For clinical practice X x For Phase II endpoint Sequencing: Can all these be accomplished in one study, as is often required by reviewers of protocols? ACRIN EISC 2008 Gatsonis, Brown University
Generalizable studies Opportunity: New imaging markers work a lot closer to the disease process and therapeutic intervention than morphologic imaging. Challenge: How granular should evaluation become? Will it be necessary to evaluate marker in every disease/therapy combination? ACRIN EISC 2008 Gatsonis, Brown University