LAUSR

INTRODUCTION In knee osteoarthritis, local increase and decrease in cartilage thickness has been observed over short time intervals. Hence, averaging cartilage change across large regions may not capture the complexity of structural alterations in disease progression. This study aims to examine the relative performance of different metrics of cartilage thickness change for different clinical studies scenarios. MATERIALS AND METHODS Metrics for assessing cartilage thickness change were characterized by conventional measures of change versus absolute values (the magnitude) of change, and by different methods of summarizing change over (sub-) regions. Sample sizes for these metrics were derived for 6-24-month observation periods, and for different treatment efficacies. Treatment effects were derived from an observational trial with 6-, 12-, and 24-month follow-up, ranging from slowing cartilage loss to stimulating cartilage growth. RESULTS Projected sample sizes ranged from 10 to >10,000 patients/arm (median = 164), depending on metric choice, treatment efficacy, and observation period. The smallest sample sizes for metrics using magnitude of change typically were half the size of those using conventional measures of change. Extreme values, e.g., minimum change or average of last four-ordered values of absolute change, required smaller sample sizes than metrics averaging over one or more regions. CONCLUSIONS Metrics using extreme magnitudes of change were most efficient in detecting differences between treatment and placebo, i.e., involved the smallest sample sizes across different DMOAD study lengths and treatment efficacies. Ancillary metrics can be used to clarify whether differences between treatment and placebo indicate structural benefit when needed.