18F-FDG PET definition of gross tumor volume for radiotherapy of non-small cell lung cancer: is a single standardized uptake value threshold approach appropriate?

UNLABELLED: PET with (18)F-FDG has been used in radiation treatment planning for non-small cell lung cancer (NSCLC). Thresholds of 15%-50% the maximum standardized uptake value (SUV(max)) have been used for gross tumor volume (GTV) delineation by PET (PET(GTV)), with 40% being the most commonly used value. Recent studies indicated that 15%-20% may be more appropriate. The purposes of this study were to determine which threshold generates the best volumetric match to GTV delineation by CT (CT(GTV)) for peripheral NSCLC and to determine whether that threshold can be generalized to tumors of various sizes.

METHODS: Data for patients who had peripheral NSCLC with well-defined borders on CT and SUV(max) of greater than 2.5 were reviewed. PET/CT datasets were reviewed, and a volume of interest was determined to represent the GTV. The CT(GTV) was delineated by using standard lung windows and reviewed by a radiation oncologist. The PET(GTV) was delineated automatically by use of various percentages of the SUV(max). The PET(GTV)-to-CT(GTV) ratios were compared at various thresholds, and a ratio of 1 was considered the best match, or the optimal threshold.

RESULTS: Twenty peripheral NSCLCs with volumes easily defined on CT were evaluated. The SUV(max) (mean +/- SD) was 12 +/- 8, and the mean CT(GTV) was 198 cm(3) (97.5% confidence interval, 5-1,008). The SUV(max) were 16 +/- 5, 13 +/- 9, and 3.0 +/- 0.4 for tumors measuring greater than 5 cm, 3-5 cm, and less than 3 cm, respectively. The optimal thresholds (mean +/- SD) for the best match were 15% +/- 6% for tumors measuring greater than 5 cm, 24% +/- 9% for tumors measuring 3-5 cm, 42% +/- 2% for tumors measuring less than 3 cm, and 24% +/- 13% for all tumors. The PET(GTV) at the 40% and 20% thresholds underestimated the CT(GTV) for 16 of 20 and 14 of 20 lesions, respectively. The mean difference in the volumes (PET(GTV) minus CT(GTV) [PET(GTV) - CT(GTV)]) at the 20% threshold was 79 cm(3) (97.5% confidence interval, -922 to 178). The PET(GTV) at the 20% threshold overestimated the CT(GTV) for all 4 tumors measuring less than 3 cm and underestimated the CT(GTV) for all 6 tumors measuring greater than 5 cm. The CT(GTV) was inversely correlated with the PET(GTV) - CT(GTV) at the 20% threshold (R(2) = 0.90, P < 0.0001). The optimal threshold was inversely correlated with the CT(GTV) (R(2) = 0.79, P < 0.0001).

CONCLUSION: No single threshold delineating the PET(GTV) provides accurate volume definition, compared with that provided by the CT(GTV), for the majority of NSCLCs. The strong correlation of the optimal threshold with the CT(GTV) warrants further investigation.