Table 2.

Strengths and Weaknesses of Five Proposed Endpoints in Chronic GVHD Therapy Trials

Proposed Endpoint	Definition	Statistical Considerations	Strengths	Weaknesses
GVHD Response	Complete plus partial response based on clinician-reported measures	• Comparison of proportions with treatment response at a specific time point	• Direct measure of success • Lengthy follow-up not needed • Easily applied	• Scales not fully qualified
Failure free survival	Survival for a defined period without new systemic treatment, death or recurrent malignancy	• Time-to-event, or • Comparison of proportions with failure-free survival at a specific time point	• Benchmarks available for 1st and 2nd-line treatment • Correlates with overall improvement reported by providers and patients • Correlates with ability to discontinue systemic treatment	• Indirect measure of failure • Improvement is not measured (i.e., GVHD manifestations may persist) • New treatment decisions are subject to bias and inconsistency • Not accepted for regulatory approvals
Survival without progressive impairment	Survival without an enduring chronic GVHD-related effect that threatens or compromises physical well-being or function in ways that cannot be easily reversed	• Time-to-event, or • Comparison of proportions surviving without progressive impairment at a specific time point	• Failure directly measured • Correlates with overall improvement reported by providers and patients	• Improvement is not measured (i.e., GVHD manifestations may persist) • Impairment is not yet fully defined • Some impairment measures might not be entirely specific for chronic GVHD • Impairment can be caused by adverse events
Patient-reported outcomes	Self-reported patient information on symptoms and multi-dimensional quality of life	• Comparison of proportions with clinically meaningful improvement at a specific time point • Comparison of distributions between study arms	• Captures the patient perspective • Lengthy follow-up not needed • Easily applied	• Subject to respondent biases • Missing data difficult to control • Claims limited to PROs
Composite scale	Selected measures from provider and patient	• Comparison of proportions with clinically meaningful improvement at a specific time point • Comparison of distributions between study arms	• Aggregates data from multiple perspectives	• Scale not developed or qualified