Variation in background intensity affects PET-based gross tumor volume delineation in non-small-cell lung cancer: the need for individualized information.

PURPOSE

Efficient tumor volume delineation by the combined use of PET/CT scanning is necessary for the proper treatment of non-small cell lung cancer (NSCLC). To understand the effect of variation in background intensity on PET-based gross tumor volume (GTV) delineation, we determined the background standard uptake values (SUVs) in normal lung, aorta (blood pool), and liver tissues and determined GTVs using different methods.

METHODS

Thirty-seven previously untreated patients with pathologically confirmed NSCLC underwent PET/CT scanning with (18)F-fluorodeoxyglucose ((18)F-FDG). To obtain (18)F-FDG uptake values in normal tissues, regions of interest in the lung lobes (left upper, left lower, right upper, right middle, and right lower), aorta, and liver zones (left, intermediate, and right) were measured. The coefficient of variation (CV) of the SUV was measured for each normal structure. The CT-based GTV (GTV(CT)) was considered as the standard to which all PET-based GTVs were compared, and the correlation coefficient was analyzed to compare GTV obtained by the various delineation methods. Linear and logarithmic regression analyses were used to determine the relationship between GTV(CT) and GTV(PET).

RESULTS

Normal lung tissue showed a significantly lower SUV and less stability than tissue of the aorta or liver. For the lung, aorta, and liver, the maximum SUV (SUV(max)) was 0.82 ± 0.32, 2.35 ± 0.37, and 3.24 ± 0.50 (CV: 38.79%, 15.82%, and 15.30%) and average SUV (SUV(ave)) was 0.49 ± 0.18, 1.68 ± 0.32, and 2.34 ± 0.36 (CV: 36.38%, 18.92%, and 15.44%), respectively. The SUVs of the lung varied from lobe to lobe. The GTV delineation method using the SUV(ave) of the lung lobe in which the tumor was found as background in the source-to-background ratio (SBR) method showed the best correlation with the volume of CT-based GTV (r=0.81).

CONCLUSIONS

Our results show vast variation in the SUV among normal tissues, as well as in the different lung lobes. The tumor volume delineated using the SBR method correlated well with the CT-based tumor volume. We conclude that it is reasonable and precise to contour GTV in patients with NSCLC after taking into account the background intensity of the lung lobe in which the tumor is found.