Comparison of lesion SUV and SUR test-retest variability in F-18 FDG PET

Quantitative assessment of radiation therapy response with FDG whole body PET has attracted increasing interest in recent years. Generally, the standardized uptake value (SUV) is utilized for quantitation. In this context, the substantial SUV test-retest variability (TRV) as demonstrated in recent studies is a major concern. The goal of our study was to investigate if standard uptake ratio (SUR) exhibits a lower TRV.

34 patients with non small cell lung cancer (ACRIN 6678 trial) were included. All patients underwent two FDG PET/CT scans prior to therapy. Maximum SUV of up to seven lesions was determined. SUR was computed as ratio of tumor SUV and blood SUV (determined in the aorta) and scan-time-corrected to 75 min p.i. TRV was assessed as the 95% confidence interval of the root mean square deviation of the paired fractional differences of SUV and SUR, respectively. The effect of blood normalization and scan time correction was inspected using the ratio RTRV=TRV[SUV]/TRV[SUR]. RTRV was correlated with the group averaged value dCFmean of the quantity dCF=|1-CF|, where CF is the factor that converts paired SUV ratios into the corresponding SUR ratios. Correlation analysis was performed by successively increasing a threshold value dCFmin and restricting the computation of dCFmean and RTRV to the respective sub-group of lesions with dCF>=dCFmin.

Overall, TRV was slightly lower for SUR than for SUV. TRV[SUR] decreases while TRV[SUV] increases with increasing dCF threshold (corresponding to larger blood SUV and scan time differences between paired scans), leading to a two-fold reduction of TRV[SUR] compared to TRV[SUV] for dCF>=15%. Correlation analysis revealed a pronounced correlation between RTRV and mean dCF (Spearman's rho=-0.99, P<0.001).

SUR is superior to SUV regarding TRV by eliminating two relevant factors contributing to TRV of SUV.