Encrypted login | home

Program Information

Quantifying Response Heterogeneity of MCRPC Lesions with NaF PET/CT

no image available
P Ferjancic

P Ferjancic1*, T Perk1 , C Lin1 , G Liu1 , R Jeraj1,2 , (1) University of Wisconsin-Madison, WI, (2) University of Ljubljana, Slovenia


TH-AB-FS4-4 (Thursday, August 3, 2017) 7:30 AM - 9:30 AM Room: Four Seasons 4

Purpose: [F-18]-NaF PET/CT has been shown to provide reliable estimate of effects of treatment on prostate cancer bone metastasis; however changes within a particular lesion have not been fully explored. This study aims to identify and quantify intra-lesion heterogeneity of response using NaF PET/CT imaging.

Methods: 80 metastatic prostate cancer bone lesions (volume > 1.5 cm³) were examined in eight patients receiving NaF PET/CT imaging during treatment with either chemotherapy or androgen receptor-directed therapy. Each patient received double baseline scans within 2-4 days and a follow-up scan after the third cycle of prescribed treatment (8 or 12 weeks). Articulated registration was used to register images and voxel-to-voxel analysis was completed on skeletal regions with best registration performance (thoracic spine, lumbar spine, sacrum and pelvis). To quantify differences within individual lesions, heterogeneity values (HV) were defined for each voxel as SUV subtracted by median(SUV) and divided by median absolute deviation (MAD) of SUV. Differences of HVs between scans were statistically analyzed within each anatomical region.

Results: HV were calculated for all lesions and compared within each ROI. MAD of HV between pelvic lesions had mean values of 0.80 (range: 0.51-0.98) for all three scans. MAD of voxel-level based changes in HV between baseline and treatment scans (ΔHVᴮᵀ) was 0.57 (range: 0.19-1.08), which was significantly larger (p=0.04) than between baseline scans (ΔHVᴮᴮ) with MAD of 0.44 (range: 0.16-1.00). Differences were also detected by KS test (p<0.001) in 14/23 pelvic lesions. ΔHVᴮᵀ distributions for several of those lesions were changes, often showing new offset peaks, indicating sub-lesional regions of different response. No significant correlation was found between lesion size and heterogeneity changes. Analysis of other anatomical regions showed similar results.

Conclusion: We quantified lesion heterogeneity on NaF PET/CT scans using HV. We showed good intra-lesional matching and detected heterogeneous response in 14/23 metastases.

Contact Email: