STANDARDIZATION OF TUMOR VOLUME DELINEATION BY PET-CT AND CT IN PATIENTS OF HEAD AND NECK CANCER-A PRELIMINARY ANALYSIS

Objectives To compare the difference in the target volumes delineated by anatomical imaging of CT scan and biological imaging with18F-FDG PET using different SUV values in terms of gross tumor volume - primary (GTVp), gross tumor volume - nodal (GTVn) and gross tumor volume - total (GTVt), and thereby determine the most appropriate threshold level for delineating tumors in FDG PET scans.

Methods 15 patients of head and neck cancers planned for concurrent chemo radiotherapy underwent a planning CT scan and FDG PET-CT scan of head and neck region. Both the data sets were then sent via a DICOM protocol to the CT simulation workstation for image co-registration and fusion. The target volumes were contoured on both separately. For PET image interpretation, focal FDG uptake was considered as positive when the activity is significantly higher than expected background structure. The tumor volumes on CT imaging were contoured according to RTOG guidelines and were labelled as CT-GTVp (primary) and CT-GTVn (nodal) and CT- GTVt (total). The tumor volumes were delineated on FDG-PET using auto contouring method taking three different threshold levels - 2.5 SUV, 30% of maximum SUV and 50% of maximum SUV and were labelled as PET-GTVp 2.5, PET-GTVp 30 and PET-GTVp 50 respectively, for nodal volumes as PET-GTVn2.5, PET-GTVn30 and PET-GTVn50, and for total volumes as PET-GTVt 2.5, PET-GTVt 30 and PET-GTVt 50. The mean of the target volumes obtained by CT and FDG PET was then calculated and compared using Wilcoxan signed rank test. Pearson’s correlation was used to determine the threshold level of FDG PET which correlated best with the CT volumes.

Results CT- GTVp (mean 45.94 cc,) was not significantly different from the PET- GTVp 2.5 (mean 44 cc), with p value of 0.49. CT-GTVn (12.28 cc) was significantly different from PET-GTVn 2.5 (8.45 cc) with p value of 0.005. However, the mean total GTV obtained by combining the volumes of primary and nodes delineated by CT (CT-GTVt = 58.27 cc) was also not significantly different from the volume obtained by PET (PET-GTVt 2.5 = 52.49 cc) with p value of 0.30. PET-GTVp 30 (26.27 cc) and PET-GTVp 50 (12.11 cc) was significantly lesser than CT-GTVp (45.94 cc), with p values of 0.01 and 0.001 respectively. PET-GTVn 30 (3.9 cc) and PET-GTVn 50 (2.07 cc) was significantly lesser than CT-GTVn (12.28 cc), with p values of 0.002 for both. The mean volume of GTV total delineated using PET with SUV 30 (PET-GTVt 30 = 30.17 cc) and PET with SUV 50 (PET-GTVt 50 = 14.19 cc) was also significantly lesser than GTV total delineated using CT (CT-GTVt = 58.27 cc), with p values of 0.009 and 0.001 respectively. PET-GTVt 2.5 was positively correlated with CT-GTVt (Pearson correlation = 0.49; p < 0.06).

Conclusions FDG PET based tumor and nodal volumes are strongly affected by the choice of threshold level. In our study, the threshold level of SUV 2.5 on FDG PET based delineation correlated well with CT based delineation, and delineation based on SUV 30% and SUV 50 % resulted in gross miss in the target volume delineation. However, a study with large number of patients is needed to justify our results further.