STatistically Assigned Response Criteria in Solid Tumors (STARCIST).

Area of Science:

• Oncology
• Nuclear Medicine
• Pharmacodynamics

Background:

• Standard FDG-PET reproducibility studies use short inter-scan intervals (1-3 days).
• Early phase drug trials assess response over longer intervals (2-3 weeks) during the first treatment cycle.
• Existing methods may not accurately capture variability in glucose uptake over clinically relevant on-treatment periods.

Purpose of the Study:

• To develop a data-driven approach for calculating baseline-specific cutoff values for FDG-PET.
• To determine patient-level changes in glucose uptake unlikely to be explained by random variability over longer scan intervals.
• To account for statistical fluctuations and potential bias effects in standardized uptake value (SUV) measurements.

Main Methods:

• Analyzed FDG-PET scans from 53 advanced cancer patients with a median 21-day interval.
• Developed a simulation-based tool using baseline SUVmax, inter-scan variability, and desired Type I error rate.
• Incorporated bias corrections for tracer uptake time variations.

Main Results:

• Changes in SUVmax followed a Gaussian distribution with constant variance, but greater variability than previously reported.
• The proposed method yielded distinct results compared to EORTC criteria for NSCLC patients treated with erlotinib.
• The method demonstrated greater statistical power to detect significant SUVmax decreases than symmetric threshold criteria.

Conclusions:

• Patient-specific, baseline-dependent cutoffs identify statistically significant changes in SUVmax based on natural glucose uptake fluctuations.
• Asymmetric cutoffs, especially for lower baseline values, require larger percentage changes for statistical significance.
• The developed method supports one-armed trials for detecting FDG-PET-based pharmacodynamic drug effects.